Title Page
 Table of Contents
 List of Tables
 List of Illustrations
 Calculations and materials
 Preliminary experiments
 Identification of products
 Comparison experiments
 Discussion of results
 Biographical sketch

Group Title: catalytic oxidation of perfluoropropene
Title: The Catalytic oxidation of perfluoropropene
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00098010/00001
 Material Information
Title: The Catalytic oxidation of perfluoropropene
Physical Description: vii, 120 l. : illus. ; 28 cm.
Language: English
Creator: Christie, Warner H., 1929-
Publication Date: 1958
Copyright Date: 1958
Subject: Catalysts   ( lcsh )
Oxidation   ( lcsh )
Chemistry thesis Ph. D
Dissertations, Academic -- Chemistry -- UF
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
Thesis: Thesis - University of Florida.
Bibliography: Bibliography: l. 116-119.
Additional Physical Form: Also available on World Wide Web
General Note: Manuscript copy.
General Note: Vita.
 Record Information
Bibliographic ID: UF00098010
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: alephbibnum - 000424009
oclc - 11069208
notis - ACH2414


This item has the following downloads:

PDF ( 6 MBs ) ( PDF )

Table of Contents
    Title Page
        Page i
        Page i-a
        Page ii
    Table of Contents
        Page iii
        Page iv
    List of Tables
        Page v
        Page vi
    List of Illustrations
        Page vii
        Page 1
        Page 2
        Page 3
        Page 4
        Page 5
        Page 6
        Page 7
        Page 8
        Page 9
        Page 10
        Page 11
        Page 12
        Page 13
        Page 14
    Calculations and materials
        Page 15
        Page 16
        Page 17
        Page 18
        Page 19
        Page 20
        Page 21
    Preliminary experiments
        Page 22
        Page 23
        Page 24
        Page 25
        Page 26
        Page 27
        Page 28
        Page 29
        Page 30
        Page 31
        Page 32
        Page 33
        Page 34
        Page 35
        Page 36
        Page 37
        Page 38
        Page 39
        Page 40
        Page 41
        Page 42
        Page 43
        Page 44
        Page 45
        Page 46
        Page 47
        Page 48
        Page 49
        Page 50
        Page 51
        Page 52
    Identification of products
        Page 53
        Page 54
        Page 55
        Page 56
        Page 57
        Page 58
        Page 59
        Page 60
        Page 61
        Page 62
        Page 63
        Page 64
        Page 65
        Page 66
        Page 67
        Page 68
        Page 69
        Page 70
        Page 71
    Comparison experiments
        Page 72
        Page 73
        Page 74
        Page 75
        Page 76
        Page 77
        Page 78
        Page 79
        Page 80
        Page 81
        Page 82
        Page 83
        Page 84
        Page 85
        Page 86
        Page 87
        Page 88
        Page 89
        Page 90
        Page 91
        Page 92
        Page 93
        Page 94
        Page 95
        Page 96
        Page 97
    Discussion of results
        Page 98
        Page 99
        Page 100
        Page 101
        Page 102
        Page 103
        Page 104
        Page 105
        Page 106
        Page 107
        Page 108
        Page 109
        Page 110
        Page 111
        Page 112
        Page 113
        Page 114
        Page 115
        Page 116
        Page 117
        Page 118
        Page 119
    Biographical sketch
        Page 120
        Page 121
        Page 122
Full Text








The author would like to express his most sincere
thanks to the many people who have made the completion of
this work possible. The author would like to give special

thanks to Dr. J. A. Wethington, Jr. His encouragement and
interest have made the undertaking of this research both
successful and enjoyable.
The personnel of the Fluorine Research Center at

Reed Laboratory has very generously provided advice and
assistance wherever this was needed for the solution of a

Members of the Chemistry Department cheerfully

provided services and equipment which greatly facilitated

the successful completion of the experimental work.

Measurements of surface areas were made by the B.E.T.

method at the Special Analytical Services Department of the

Oak Ridge Gaseous Diffusion Plant.
The sample of RbF used in this work was supplied by
the Oak Ridge National Laboratory.
This work was supported by the Office of Naval



. . ii



* U S U U U U


* .

. . vii




Apparatus Used in tl
Apparatus Used in tl
Distillation System
Analytical System

he Preliminary

he Comparison


Chemicals Used


Experiments Over CuO and Hopcalite
Experiments Performed Over NaF
Experiments Performed Over Nickel Packing
Experiments Performed Over CsF
The Attempted Oxidation of CF3CF2CF3 Over
Determination of the COF2, CF3COF Ratio
Some Conclusions Drawn from the Preliminary


* U U U U


* U


ACKNOWLEDGMENTS . . .. . . . . .


Identification of CF3C(OH)2CF3
Identification of COF2
Identification of CF3COF
Identification of CF3CF2COF
Identification of CF3COCF3
Identification and Proof of Structure of
Identification of CF3COCF2CF3
Tentative Identification of (CF3)2CFCOCCF2C3
Piperazine Derivatives of Some Perhalo-Acids

General Procedure Used in the Comparison
Comparison Experiments Using LiF
Comparison Experiments Using NaP
Comparison Experiments Using KF
Comparison Experiments Using RbF
Comparison Experiments Using CsF
Comparison Experiments Using Nickel
Comparison Experiments Using CaF2
Comparison Experiments Using BaF2
VIII. CONCLUSIONS . . . . . . . . 114


LITERATURE CITED . . . . . . . . .



1. Reactions of CF3CF:CF2 over CuO and Hopcalite . 23

2. The Reaction of CF3CF:CF2 with 02 over NaF . 31

3. Distillation of Products Obtained in Run 134 I 34

4. Distillation of Products Obtained in Run 33 II 36
5. The Reaction of CF3CF:CF2 with 02 over Nickel . 38

6. The Reaction of CF3CF:CF2 with 02 over CsF . 42

7. Distillation of Products Obtained in Run 52 III 46
8. Determination of Mole 7o COF2, in the COF2 and
CF3COF Mixtures, for Some Preliminary
Experiments . . . . . . 49
9. Infrared Spectrum of CF3C(OH)2CF3 . . . . 55

10. Molecular Weight Determinations on COF2 . . 58

11. Infrared Adsorption Bands for CF30CF2COF . . 63

12. Piperazine and N-Phenylpiperazine Derivatives
of Some Perhalo-Acids . . . . . . 68

13. Reactor and Catalysts Data, Comparison
Experiments . . . . . . . . . 77

14. Data from Comparison Experiments over LiF . . 78

15. Data from Comparison Experiments over NaF . . 81

16. Data from Comparison Experiments over KF . . 83
17. Data from Comparison Experiments over RbF . . 87
18. Data from Comparison Experiments over CsF . . 90



19. Data from Comparison Experiments over Ni,
CaF2 and BaF2 . . . . . . . . 93

20. Surface Area Data for the Reactors Used in the
Comparison Experiments . . . . .. 1 00

21. Melting Point Correlations . . . . . 103


Figure Page

1. System Used in the Comparison Experiments . 12

2. Comparison Experiments Using LiF . . . 79

3. Two Sets of Comparison Experiments Using NaF 82

4. Comparison Experiments Using KF . . . . 84

5. Comparison Experiments Using RbF . . . 86

6. Comparison Experiments Using CsF . . . 89

7. Comparison Experiments Using Ni . . . 94

8. Comparison Experiments Using CaF2 . . . 95

9. Comparison Experiments Using BaF2 . . . 96

10. Summary Plot; Percentage Conversion as a
Function of T,oC for the Alkali Fluorides . 99

11. Percentage Conversion as a Function of
Absolute Temperature . . . . . . 102

12. Summary Plot; Percentage Conversion as a
Function of T,oC for CaF2, BaF2 and Ni . 106




The phenomenon of catalysis was first observed and
recorded in 1835, by J. J. Brezeliusl who introduced the term
catalytic agent into chemistry to include those substances
which, by their mere presence and without being altered
themselves, accelerate the rate of reactions. In reversible
systems the catalyst accelerates the establishment of
equilibrium until at the equilibrium point both forward and
reverse reactions are equally accelerated. It is possible to
distinguish between homogeneous and heterogeneous catalytic
reactions. The former generally occur in gases and liquids
where the catalyst is dissolved, and the latter take place at
gas-liquid, gas-solid or liquid-solid interfaces, the
catalyst usually being the liquid or solid.
The theoretical chemist has attempted to understand
the phenomenon of catalysis by way of the kinetic approach.
This approach has not proved altogether satisfactory.
Basically, the method of chemical kinetics involves the quanti-

tative determination of the reaction products and their
dependence on time and temperature and the expression of

rates in terms of concentrations and temperatures. A model
or mechanism is then constructed to fit the observations.
Such a mechanism has to include assumptions regarding the

step or steps involved in the catalysis proper. A mechanism
deduced in this fashion is generally considered correct until
more or better measurements prove it wrong. It is not
surprising that mistakes are made in interpreting kinetic
data when it is considered that at least five steps2 are
involved in a heterogeneous catalytic reaction. These are
the diffusion of the reacting molecules to the surface, the

adsorption of the gases on the surface, reaction on the
surface, desorption of the products and finally the diffusion
of the desorbed products into the main body of the gas
The slowest of these steps will determine the rate of

the over-all reaction. In many cases wrong conclusions have

been drawn as to which step is actually rate determining in a
given reaction. It is no surprise, therefore, that the
mechanism of the rearrangement of molecules on a surface
cannot be deduced from kinetic measurements alone. A
systematic approach would use kinetic data to supplement more
direct physical measurements and observations concerning the
surface involved.
Recently, investigators have shifted their attention
from the reactants and products to a detailed consideration

of the structure and properties of the solid catalyst itself.
The techniques and theories of solid state physics have now
been successfully applied to a number of solid catalysts.
Consequently, at this time, the electronic nature of the
solid has become a factor of major interest to workers3'4
in this field.
The catalyst unit may be thought of as a group of
near neighbor atoms whose configuration or properties are
such that they exert the required forces on the reactant to
assist in breaking some bonds and forming other bonds.
Proper orientation of the reactants is required on the surface
so that the atoms needed for formation of products are in
close proximity. The lattice spacing or the geometry of the
surface of a solid and the unsaturated valence forces
determine the forces of attraction of different catalyst atoms
for the various parts of the reactant molecules. As such
forces only operate over short distances, the interaction is
largely determined by the relative spacing of surface and
reactant atoms. The discovery of fivefold differences in the
intrinsic catalytic activity of oriented and non-oriented
films of the same metal proves conclusively the importance of
the catalyst lattice parameter in catalysis.5
The purpose of this investigation was to attempt to
find materials which would serve as catalysts for the
reactions of fluorocarbons. Saturated flurocarbons are

known to possess great chemical and thermal stability. These

properties make fluorocarbons extremely valuable in certain
applications but also make it impossible to apply the
classical synthesis or degradation methods used in organic
Fluorocarbons as a field of chemical substances were
only first reported in 1937. In that year and more
extensively in 1939 the preparation and properties of liquid
fluorocarbons were disclosed by Simons and Block.6,7 Because
of certain desirable properties, fluorocarbons were in demand

during World War II. The science and technology of fluoro-
carbons was still in its infancy and the impetus of war
spurred many developments in the field. Several methods were
devised for the production of fluorocarbons. A modification

of the catalytic vapor-phase fluorination technique introduced
by Fredenhagen and Cadenbach8 and by Bigelow and co-workers9-12
was developed at Columbia University. At the same time
the metallic fluoride process13 was being perfected at the
Johns Hopkins University. In this same period still another
method was invented by Simons.14 This was the electrochemical
process the announcement of which appeared in 1949. The
Simons method has proved to be the most versatile for the
production of fluorocarbons.
Since the science of fluorine chemistry is relatively
new, few guiding principles were available for the selection
of materials to be used as catalysts. The approach used

in this work is one familiar to those working in catalysis.
Substances that could be conceived of as possible catalysts
were simply tested, in an experiment, to determine whether
or not they were suitable.
The reaction of CF3CP:CF2 with oxygen was investi-
gated because this system offered a definite possibility for
formation of useful products. Fluorocarbon olefins are known
to be reactive. The simultaneous oxidation and hydrolysis of
olefins containing a perfluoro-alkyl group has been used for
the preparation of acids.15
3CF3CC1:CC12 + 4K1n04 + 14KOH

3CF3CO2K + 4MnO2 + 3K2C03 + 7H20 + 9KC1.

In a similar fashion the oxidation of'(CF3)2C:CF2 produces
hexafluoroacetone in good yield.16 Highly'unsaturated
perfluoro-olefins such as CF2=CF-CF:CF2 are reported to
undergo oxidation17 in air or oxygen. Swarts has reported
the oxidation18 of CFBr:CF2. The combustion of CF2:CF2 in
oxygen is reported by Benning and Parks19 to result in the
formation of CF4, although, in general, acid halides appear
to be the principal products formed by the air oxidation of
perhalo-olefins. Hurka prepared CClF2COF in good yield by
oxidizing20 CCFICF2 with oxygen in the absence of a
catalyst in an autoclave at 100-300 lb. per sq. in.
Myers21 reported the formation of a peroxide in the
oxidation of CC1F:CF2 with 02. These experiments were

performed at room temperature, at autogenous pressures, and
in the absence of light. The peroxide formed was never
isolated in the free state.
A number of attempts at oxidizing CF3CF:CF2 over
some of the classical catalysts of organic chemistry failed
to give satisfactory results. Work done by Gens and
Wethington suggested that alkali fluorides might serve as
catalysts for this reaction. In a recent paper22 these
workers demonstrated that certain fluorocarbon compounds,
CF3CPFCF2 in particular, exchanged fluorine atoms rapidly
with the alkali fluorides. The order of exchange was
reported to be Cs )Rb >K, Na, Li.
The work done in this research is divided into two
phases. The first phase is entitled "Preliminary Experiments,"
and in this section there are described the various experi-
ments performed in the search for materials that would
function as catalysts for the oxidation of CP3CF:CF2. The
second phase is entitled "Comparison Experiments." In this
section a set of experiments is described in which various
surfaces are evaluated, in a quantitative fashion, for
ability to catalyze the oxidation of CF3CF:CF2.



Apparatus Used in the Preliminary Experiments

A reactor tube was fabricated from a 33 cm. length

of 2.14 ca. I.D. nickel tubing. The ends of this tube were

closed by attaching nickel plugs with silver solder. Each

plug had a 3/8 in. outlet hole onto which a 15 ca. length

of 3/8 in. nickel tubing was flush silver soldered. A

second reactor tube, 15 ca. in length, was constructed in
the same fashion.
The reactor temperature was' controlled by a
Simplytrol pyrometer unit which automatically turned the
furnace unit off and on to maintain, within t5, the prese-
lected temperature. An iron-constantan thermocouple, silver
soldered to the center portion of the reactor, actuated the

pyrometer unit. The reactor was heated by a clam type tube
The gaseous materials to be reacted were passed
through the reactor in either of two ways. First, the
materials were taken directly from the storage cylinders and

were metered into the reaction system in the desired ratio.

Secondly, the materials were mixed in the desired ratio in a
27 1. glass container and were then displaced by water, at
some given rate, into the reaction system.
Before entering the reaction system, the wet gases

.passed through two drying tubes. The first tube which

contained Drierite was 90 cm. in length and 2 cm. in
diameter. The second tube which contained P205 was 40 cm.
in length and 2 ca* in diameter. The reaction products were
collected in a suitable series of cold traps. The end trap

was protected from moisture by a drying tube.

Apparatus Used in the Comparison Experiments

Individual reactors were fabricated for each catalyst
studied in the comparison experiments. These reactors were
constructed in the same fashion as these used in the
preliminary experiments. The reactor dimensions were:
length, 15 ca.; I.D., 2.14 cm.; with outlet tubes 25 cm. long.
Iron-constantan thermocouples were silver soldered to the
center of each reactor tube. This thermocouple was connected
to a potentiometer that was capable of reading 0.001 mv. The
average temperature range observed with this potentiometer
is the value reported in the tables in the comparison
experiments. Temperature control was maintained by a
Simplytrol pyrometer. The controlling thermocouple was
placed in a nickel well of 7.0 ca. length, and 0.3 ca. I.D.,
which was attached to the center part of the reactor tube

by wrapping with asbestos tape. The entire reactor tube was
then insulated by wrapping it with several layers of
asbestos tape. The pyrometer unit operated the furnace
through a variable Powerstatt. The Powerstatt was operated
at approximately 15 volts above that required to maintain
the desired reactor temperature. This arrangement was found
to be capable ef holding any temperature up to 4000 within
A number of glass product traps were prepared with
stopcocks on each outlet tube. These traps were used to
collect and weigh the reaction products. Each trap was
approximately 20 ca. in length by 2.5 cm. in diameter. The
center outlet tube came within 3 cm. of the bottom of the
trap. This provision was made so that all condensable gases
would be pulled through the cold zone when the traps and
products were being degassed.
A schematic diagram of the flow system used is shown
in Fig. 1. The reactant gases were contained in a 27 1.
water operated bottle, A. The feed material was metered
through a needle valve, B, which was connected to a wet test
meter, C. The water saturated reactants then passed through
two drying tubes. The first drying tube, D, was 90 cm. in
length by 2 cm. in diameter and contained indicating
Drierite. The second drying tube, E, contained P205 and was
50 cm. in length by 4 cm. in diameter. These drying tubes
were connected to a Fischer and Porter Flowrator, F, which

had been calibrated for the feed mixture used. This meter
was connected to a three way stopcock, G, which connected to
the reactor, H. The third opening on stopcock G was open
to the atmosphere through two drying tubes, I and J. Upon
completion of a run, dry nitrogen was swept through the
reactor from stopcock G through drying tubes I and J.
The outlet of the reactor, H, was connected to another
three way stopcock which could be opened to the atmosphere
and the issuing gas stream checked with indicating paper to
ascertain whether or not reaction was occurring. Stopcock
K was connected to the product receiver, L, which was open
to the atmosphere through a drying tube, M. All tubing
connections were made with new, 3/8 in. Tygon tubing.

Distillation ystem

A distillation column was used to separate the
products formed in the oxidation reactions. This column was
of vapor take-off design and would handle materials boiling
as low as ~600. For ease of maintenance, the column was made
in three separate pieces each of which was fitted with
standard taper joints. The head had a Dewar type container
built in, which could be cooled with Dry Ice-acetone, to act
as a condenser surface. The column tube was 45 ea. in
length with an I.D. of 0.8 cm. and had a silvered vacuum
jacket. This column was packed with 1/16 in. I.D. single-
turn, stainless steel helices of No. 36 B and S gauge wire.


A. Water Operated 27 1. Storage Battle
B. Needle Value
C. Wet Test Meter
D. Drierite Drying Tube
E. P205 Drying Tube
P. Flowrator
G. Three Way Stopoock
H. Nickel Reactor Tube

I. P205 Drying Tube
J. Silica-Gel Drying Tube
K. Three Way Stopcock
L. Product Receiver
M. Drying Tube
N. Nitrogen Tank


Reactor, H


Fig. I, System Used in the Comporison Experiments

For hydrocarbons, this column was estimated to have the
equivalent of 30 theoretical plates.23 For fluorocarbons,
it was believed that this value would be reduced somewhat.
The distillation pots, of various sizes, were equipped with
thermocouple wells and were constricted at the bottom to
accommodate a heating coil.
The colaun head was permanently attached to an all
glass vacunu system specially constructed for use with a
distillation column. This vacunu system had a series of
four traps to accommodate the various fractions of a
distillation, a manometer to measure pressure, a P205 tube
for drying, and an outlet for making Regnault24 type molecular
weight determinations.

Analytical System

An analytical system was devised to determine the
percentage of COF2 in the COP2, CF3COF mixture obtained in
some of the preliminary experiments. This same apparatus was
used to determine the percentage conversion values listed in
the comparison experiments.
A 500 al. filter flask was fitted with a wide mouth
inlet tube, through a rubber stopper, that extended to
within 1 cm. of the bottom. The external end of the inlet
tube was equipped with a stopcocki The rubber stopper was
also fitted with a separatory funnel. This apparatus was
tested and found to be leak proof when assembled. A glass

enclosed iron stirring bar, which was operated by a powerful
external magnet, was included in the flask. The side arm
outlet on the filter flask was connected to a Drierite drying

tube which was connected in series to a P205 drying tube.
The P205 drying tube was in turn connected to a stopcock
equipped product trap. In operation, enough 20 per cent
NaOH solution was placed in a flask to cover the inlet tube
with 2 to 3 cm. of solution. One pass through this solution,
with vigorous stirring, was found to effectively remove all
the acidic products. Carbonyl fluoride was determined by
acidifying the basic solution and measuring the CO2 evolved.
This was accomplished by closing off the inlet tube stopcock
and adding concentrated H2S04 from the separatory funnel
while stirring the solution.



The contact times reported in some preliminary
experiments and in all the comparison experiments were
calculated from the following equation:
C. T. (- W' (1)

where C. T. average contact time,
Mt a moles of gas mixture passed through the
reactor in time t,
Mr moles of gas mixture contained in the reactor
at the reaction temperature,
t time required for Mt moles of gas mixture to
have passed through the reactor.
Applying the ideal gas law, Mr and Mt were expanded
to the following:

Mr -=

and Mt W-() (

where P atmospheric pressure (assumed to be 760 a.m.

for all experiments),

Vr free volume of the packed reactor,
R gas constant,
Tr reaction temperature, oK,
V gas volume passed in time t,
T = room temperature.
Substituting these expressions for Mr and Mt back
into equation (1) gave equation (2):

C. T. (Vr) (T) (t) (2)
(Tr) (V)
The measured volume of gas mixture passed, Vm,
required a correction for the water vapor removed by the
drying tubes to give the true volume of gas mixture passed,
V. This correction took the following form:

Vm = V + VH20, where (3)

VH20 volume of water vapor removed.

Application of Boyle's law to the term VH20 gave the
following relation:

VH20 8HO'm (Vm), where (4)

PH20,m.m. vapor pressure of H20 at room temperature.

Substitution of (4) into (3) and rearranging gave
equation (5):

V = Va- H20 (Va). (5)

Substitution of (5) into (2) gave the final form of
the contact time relation used for calculation:

C. T (Vr) (T) (t)
Tr VW- H20 z*L. (Va)

The term Vr was calculated from the following
Vr VE.R.-Vc ,
where VE.R' calculated volume of the empty reactor,

Vc a volume of catalyst in reactor.
The quantity Vc was evaluated by dividing the
weight of catalyst charge by its density. No information
was found in the literature concerning the variation of
density of the solid alkali fluorides with temperature. To
overcome this difficulty the following approximation was
made. Density values reported in Galins Handbuch de
anoranischen Caij25-29 were plotted versus temperature.
Unfortunately these reported values fell into two classes,
those determined at or near room temperature and those
determined at or near the melting point of the salt. For
each salt these points were connected with a straight line.
The average density value for the temperature range of each
set of experiments was determined from the graph for each
alkali fluoride. This average density value was used to
calculate Vc.

For the two Group II fluorides used the only
reported density values found30,31 were those at room
temperature. Consequently these are the values that were
used in the calculations of Vc. These values introduced a
small error into the calculated contact times for the
experiments with CaP2 and BaF2.
In a strict sense the tera Mr should be replaced by
an expression that accounts for the amount of reaction
taking place that would change the number of moles of
material in the reactor. Although this factor probably
influences the contact time values at high percentage
conversions, no attempt was made to correct for it. The
error introduced by this effect would probably be smaller
than the variation in the contact time.

Chemical 1sed

Hopealite oxidation catalyst was purchased from the
Mine Safety Appliance Company. This catalyst is generally
a mixture of MnO2, CuO, Co203 and Ag20.32 No analysis was
available for this material. The product was supplied as a
14/20 mesh granular solid and was used directly in this form.
Cupric oxide, "Baker and Adamson" reagent grade, was
supplied in wire form. No appreciable impurities were
present. No treatment other than streaming with oxygen at
4000 was performed on this oxide before using.

Lithium fluoride, "Baker Analyzed" reagent grade,
was supplied as a fine powder. This material was unsatis-
factory in this form since it produced an excessive pressure
drop across the reactor. This condition was alleviated by
preparing a mixture of the insoluble powder with water and
taking the resulting slurry to dryness on a hot plate while
stirring. After screening out the fine material and crushing
the large lumps, the product obtained was found to be suit-

able for use in the reactor. The chief impurities in this
fluoride were listed as K, 0.01% and Na, 0.07%.
Sodium fluoride, technical grade, was obtained from
the Harshaw Chemical Company as 1/8 in. tablets. No
analytical data was available for this material. Before
these tablets were used, they were crushed and the fine
particles were eliminated by screening.
Potassium fluoride, "Baker and Adamson" reagent
grade, was supplied in granular form. The major impurities
were listed as HP, 0.05% and K2CO3, 0.10%. This granular
form was found suitable for use in the reactor without
further treatment.
Rubidium fluoride, in finely powdered form, was
obtained from the Oak Ridge National Laboratory. This
powder was unsuitable for use in the reactor. To produce a
coarse material, this fluoride was recrystallized from
water in a platinum vessel by taking the solution to
dryness on a hot plate. It was necessary to use platinum

vessels rather than glass as this fluoride is sufficiently
hydrolyzed in aqueous solution to attack glass. This
fluoride is also reported to be decomposed by CO2 in
aqueous solution to form HF which would attack glass. The
chief impurities in this material were reported to be K,
1.03% and Cs, 0.16%.
Cesium fluoride was obtained from the A. D. Mackay
Company. The reported impurities included 0.12% Rb. This
powder was recrystallized from water, in a platinum vessel,
to obtain a form more suitable for use in the reactor.
Barium fluoride, "Baker and Adamson" reagent grade,
was used in the comparison experiments without further
"Baker and Adamson," reagent grade CaP2 was obtained
as a fine powder. This powder was slurried with water and
taken to dryness on a hot plate to produce a product
suitable for use in the reactor.
Perfluoropropene was prepared by the decarboxylation
of CF3CF2CF2COONa according to the procedure of La Zerte.33
This material was scrubbed twice through 20% NaOH to remove
any CO0 present. Molecular weight checks on the final
product were always within 1% of the theoretical value.
The piperasine and N-phenylpiperazine derivatives of
a number of perhalo-acids were prepared in the course of this
research. The following perfluoro-acids, CF3(CF2)nCOOH
(where n 0,1,2,4 and 6) and HOOC(CF2)3COOH were obtained


from the New Products Division of the Minnesota Mining and
Manufacturing Company. The following perhalo-acids,
C1(CF2CFC1)2 CF2COOH and C1(CF2CFC1)3COOH were obtained from
the Chemical Manufacturing Division of the M. W. Kellogg
Company. No analyses were specified for these acids which
were used in the preparation of derivatives without further



Reference is occasionally made to low-boiling and
high-boiling products. This usage indicates, respectively,
that the material being discussed boils below or above the
fluorocarbon starting material. All temperatures referred
to in this manuscript are on the Centigrade scale unless
otherwise stated.

Exe;riments over uQ ad Hopcalite

Four preliminary experiments were performed with
CuO. Quantitative data concerning product yields is lacking
but useful qualitative deductions can be made from these
runs. These experiments are best summarized in Table 1. A

brief paragraph is included about each run in the following
Experiment 65 I: The reactants were metered
separately into a 33 cm. reactor packed with 331 g. of CuO.
The center part of this reactor was maintained at 5060. One
hundred and nineteen grams of material were collected in
the Dry Ice and liquid air-cooled product traps. By weight,



Mole Flow Rate
Run No. Reactor Reactor Ratio cc./min. Moles of CF3CCFCF2 Products and Remarks
and Tep. O2/CF3 at Room
Packing CFCF2 Temp. Initial Recovered Used
Reactions Over CuO
65% by weight of material
33 cm. believed to be CO2. 6.0 g.
65 I Reactor 5060 0.10 28.2 0.71 0.25 0.46 of neutral high boiling
331 g. material. Trace of

33 c*m Recovered 90.5% of start-
89 I Ractor O Olefin 6.3 0.35 0.31 0.0 ing material. Some CO2
u Only obtained.

Recovered 69% of starting
93 I Same as 450o Olefin 8.1 0.24 0.17 0.07 material. 14l by weight
89 I Only C02 in products.

Recovered 35% of starting
96 I Same as 300 0.50 9.9 0.0 0.1 0.26 material. Large amount of
93 1 C02 formed. Reactor
plugged by CuF2 formation.

TABLE 1 Continued

Mole Flow Rate
Ran No. Reactor Reactor Ratio cc./min. Moles of CF3CF-CF2 Products and Remarks
and Temp. 02/CF3 at Room
Packing CF-CF2 Temp. Initial Recovered Used

Reactions Over Hopcalite
15 cm. Obtained 87% conversion to
119 Reactor 300 0.53 9.6 0.23 0.03 0.20 CO2. No other products
119 I 62 g. ound.
Hopcalite ound

Reactor Obtained 35% conversion to
120 I 63 g. 2000 0.57 13.0 0.27 0.09 0.18 CO CFC(OH)2CF3 made in
Hopealite 3. yi d.

15 cm.
122 Reactor 150-1800 0.98 23.0 03 0.26 0.08 Recovered 77 of olefin
I 61 g.o un ctd

65% of these products were soluble in NaOH solution. Absence

of fluoride ion in this solution indicated that this
soluble material was CO2. Six grams of material boiling
above 00 was obtained. Chromatographic analysis indicated
that seven components were present in this material.
Approximately 0.5 g. of white, needle like crystals were
obtained in the wet ice-cooled trap. This material was
found to be CF3C(OH)2CP3.
Experiment 89 I: The reactor used in the previous
run was repacked with fresh CuO for this experiment. Pure
olefin was passed through the reactor at 4000. On
distillation, 90.5% of the starting material was recovered.
A small trace of both low-boiling and high-boiling material
was obtained. In view of these results the experiment was
repeated at a higher temperature.
Experiment 93 I: The reactor used in experiment
89 I was used again without change in this run. Pure
olefin was passed through the reactor at 4500. Upon distil-
lation 69% of the starting material was recovered.
Approximately 14% by weight of the products was CO2. A
trace of high-boiling material was obtained.
Experiment 96 I: The reactor used in the previous
experiment was used unchanged in this run. Oxygen and
CP3CF:CF2 were mixed in the 27 1. storage bottle in a one to
two mole ratio. These gases were metered into the reactor
at 4300. Distillation showed that 35% of the starting

material was unchanged. The material coming off overhead
was taken up by Ba(OH)2 solution. The precipitate thus
obtained was completely soluble in dilute HNO3 with the
liberation of CO. This indicated the absence of material

containing hydrolyzable fluorine. Upon completion of this
run the CuO in the reactor was inspected. The formation of
red CuF2 explained the poor weight balances noted in these
Results obtained in the oxidations over CuO were not
satisfactory as the yield of useful products was very low.
CuO appeared to be a reactant in this system rather than a
Three preliminary experiments were performed with
Hopcalite. This catalyst was found to be more active than
CuO. These three experiments are tabulated in Table 1. The
significant details, pertaining to each run, are discussed
in the following paragraphs.
Experiment 119 I: The 15 cm. reactor was packed
with 62.2 g. of commercial Hopealite catalyst. Thirty-four
grams (0.23 mole) of CF3CFPCF2 were mixed with 3.9 g.
(0.13 mole) of oxygen gas in the water operated storage
bottle. This material was passed through the reactor, held
at 3000, at a rate of 9.6 cc. per min. Of a total of 38 g.
of material passed, only 29 g. were recovered. The catalyst
had changed color and was white and brown in places indicating
that it had participated in a reaction with the olefin.

A molecular weight determination on the first
fraction of the products gave a value of 44.7 indicating the
presence of CO2. The entire gaseous material was scrubbed
twice through a NaOH solution to yield 4.5 g. of unreacted
olefin. A check on the solution contained in the scrubber
revealed the absence of fluoride ion. The apparent reaction
CF3CF:CF2 + Metal oxide 3002 + Metal fluoride.

On this assumption it was calculated that 30.5 g. of material
should have been recovered which was in fair agreement with
the 29.0 g. recovered. It was assumed that the oxygen
originally mixed with the olefin did not react at all and
escaped from the product trap. Four and one-half grams of
olefin were recovered indicating that 29.5 g. (0.20 mole)
reacted to form CO2. Thus the recovered material should
have consisted of (0.20) (44) (3) 26.0 g. of CO2 and 4.5 g.
of olefin for a total of 30.5 g.
Since oxidative degradation took place to such a
large extent a similar experiment was performed at a lower
Experiment 120 I: Forty-one and one-half grams of
CF3CF:CF2 (0.28 mole) were mixed with 5.1 g. of oxygen
(0.16 mole) in the water operated storage bottle. The 15 cm.
nickel reactor was refilled with fresh Hopcalite. This
material was passed through the reactor held at 2000 at a
flow rate of 13 cc. per min. Of a total of 46.6 g. of

reactants used, only 36.1 g. of material were recovered
from the product traps. This material was transferred into
the distillation column. One and six-tenths grams of a
white solid were left behind as a residue in this transfer.
This solid was CF3C(OH)2CF3.

The material in the distillation column was distilled
to give 16.9 g. of CO2, 13.0 g. of unreacted olefin and
6.1 g. of material boiling above -27. The results of the
distillation can be treated as follows. The 16.0 g. of

C02 obtained would account for (16.0/44) (1/3) 0.12 mole
of olefin oxidized. The 13.0 g. of CF3CP.CF2 recovered
accounts for 0.09 mole of olefin. The white solid accounts
for 0.01 mole of olefin using 182 as its molecular weight.
This leaves 0.060 mole of material to be accounted for by the
6.1 g. of material boiling above -27o. Molecular weights on
this 6.1 g. ranged from 152 to 164. Assuming an average
value of 158, 0.04 sole of olefin were accounted for. This
mole balance was in good agreement with the observed results.
Moles of olefin used initially equaled 0.28. Moles of
olefin accounted for equaled 0.26. These two experiments,
119 I and 120 I, indicated that the major process occurring
over Hopealite was:
CP3CCFCF + Metal oxides 3CO2 + Metal fluorides.
It is believed that the formation of CF3C(OH)2CF3,
rather than CF3COCF3, was caused by traces of moisture in

the reaction system. Possibly small amounts of CF3COCF3
were also formed but this compound was not detected in the
free state. Since CF3C(OH)2CF3 was formed in 3.5% yield
another experiment at a lower temperature was made in an
attempt to produce the anhydrous ketone as well as the
hydrated form.
Experiment 122 I: Fifty-one grams of CFSCF:CF2
(0.34 mole) were mixed with 10.7 g. of oxygen (0.33 mole)
in the water operated storage bottle. The same 15 cm.
reactor repacked with fresh Hopcalite was used. The
reactants were passed through the reactor held at 1500 at a
flow rate of 23 cc. per min. for 260 min. The temperature
was increased to 1800 for the remaining 480 min. of the run.
Out of a total of 61.7 g. of materials used only 41 g. were
recovered from the product traps. This material was
transferred into the distillation column. No very low
boiling material (CO2) was isolated as in the previous run,
120 I. Thirty-nine grams of unreacted olefin, representing
77%, were recovered. Approximately 2.0 g. of a water
insoluble, bluish liquid remained as a pot residue after
distillation. The poor material balance observed in this
run could be explained by the formation of copper or
manganese salts of fluorocarbon acids which could be stable
at the temperature of the run. LaZerte, et a .,33 reported
that (C3F7CO)2 Cu decomposes at 2900-300.

Results obtained in the oxidations over Hopcalite
were unsatisfactory as were those over CuO. Useful products
were not formed in appreciable quantities. The major
process was found to be reaction of the Hopcalite with
CFP3CFCF2 to give CO2.

Experiments Performed QOve NaF

Eight preliminary experiments were performed using

NaP packing in the 33 cm. reactor. These experiments are
tabulated in Table 2. In each run the reactants were
premixed in the 27 1. water operated storage bottle to give
the values reported in Column 2. Contact times listed in
Column 3 are only average values as precise flow control was
not maintained. In each experiment the products obtained
were separated by distillation. Percentage conversion
values, Column 7, were calculated from the ratio of Column 6
to Column 4. Percentage by weight of COF2 and CF3COF in the
products, Column 8, was calculated from the total weight of
products after the amount of recovered olefin was deducted.
This value, for NaF, does not seem to be greatly affected by
the varied operating conditions used. Since these experiments
were all performed in a similar fashion only two
representative runs will be discussed in detail.
Experiment 134 I: Eighty-five grams (0.57 mole)
of CF3CF:CF2 were mixed with 16.8 g. (0.52 mole) of oxygen
in the 27 1. storage container. The reaction mixture was


Rector Mole Ratio Approx. Moles of CF3CF=CF2
Run No. Temp. O/CCFrCF Contact Time
T 02/ C2 in Seoands Initial Recovered Used

(1) (2) (3) (4) (5) (6)
110 I 3750 0.50 248 0.35 0.09 0.26

113 I 3500 0.62 210 O.44 0.17 0.27

132 I 3150 0.85 47-136 0.53 0.53

134 I 2800 0.93 67 0.57 0.01 0.56

135 I 2000 1.05 168 0.23 0.23 -

136 I 2500 1.02 76 0.22 0.11 0.11

33 II 2800 1.10 52 0.22 0.22

6 III 3010 1.1 0.80 0.80

TABLE 2 Continued

% by Weight % by Weight
% Conversion of COF2 and of Products Remarks
to Products CF30F in Boiling Above
(7) (8) (9)
74 65 35 CF3CF~ag in 18.5% yield.

61 75 25 CF3CF2COF in 20.0% yield.

00 70 CF3CF2COF in 15.5% yield.
SCF30F2COF trace.

98 63 37 CF3CF2COF in 20% yield.
CF3OCF2COF in 6% yield.

0 No reaction

No yields determined on
0 63 37 products boiling above
So 63 37 -26.80. Quantity too
small to fractionate.

100 76 24 CF3CF2COF in 9% yield.
CF30CF2COF in 12% yield.

100 CF30CF2COF in 10% yield.

passed through the 33 cm. reactor held at 2800. The material
collected in the traps amounted to 106 g. This material
was distilled to give the results tabulated in Table 3.
After removal of the material boiling below -28.2,
two distinct flats were observed in the distillation curve.
The product distilling from -28.2o to -26.70 was mainly
CF3CF2COF with a trace of CF3COCF3 and unreacted CF3CF:CF2
present. The material from -10.8 to -9.60 was a
previously unreported compound. This new acid fluoride was
identified as CF3OCF2COF and gave rise to a new fluorocarbon
acid containing an oxygen linkage. The unidentified pot
residue was found to be water soluble and produced an
acidic solution.
Product yields were estimated to be CF3CP2COF,
19.7%, CF3COCF3, 1%, and CF30CP2COF, 6%. Based on the ratio
of two moles of COF2 to one mole of CF3COF, produced by
equations (6) and (7), the yields of these products were
estimated to be 62.5% COF2 and 46.4% CF3COF.
Estimating the yields of COF2 and CF3COF was
complicated by the fact that two moles of COF2 were generated
for each mole of CF3COF obtained. The following reaction
scheme could possibly explain this:

CF3CF:CF2 + 02 CF3COF + COF2 (6)

CF3CF:CF2 + 102 3COF2.




Possible Material


Overhead 50.5 up to -57.70 COF2, CF3COF 66.8-117

FCOFF, trace of
1 14.2 -57.7 to -28.2 CF 117-152

CF3CF2COF, trace
2 18.9 -28.2o to -26.70 of CF3COCF3 and 155-167

3 9.9 -26.70 to -10.80 Intercut 167-180

5.8 -10.80 to -9.6o CF30CF2COF 180-182

Residue 4.0 -9.60 and up 182-282


in Orans



It was assumed that the excess COF2 above that produced by
equation (6) was produced by equation (7). This scheme is
feasible since a material balance indicates that the
additional COF2 cannot be accounted for by oxidizing terminal
CF2 groups.
Out of 0.56 mole of CF3CFPCF2 consumed, 0.11 mole
was converted to CF3CF2COF and 0.03 mole was converted to
CF3OCF2COF. Approximately 0.06 mole was accounted for by
Cut 3, Table 3, assuming this material was mainly three
carbon products of average molecular weight of 174. Reaction
(6) accounts for another 0.26 mole of olefin. All of these

products accounted for a total of 0.46 mole of converted
olefin. Only 0.11 mole of terminal CF2 groups was left.
This quantity did not account for the 0.26 additional mole
of COF2 formed, hence equation (7) is believed to be correct
as this equation requires only 0.09 mole of olefin to account
for the 0.26 mole of COF2 found. Formation of COP2 by this
path would account for the remaining 0.11 mole of olefin.
The agreement in mole values noted here was well within the
expected error in experimental technique.
Experiment 33 II: Thirty-two and eight-tenths grams
(0.22 mole) of CF3CF:CF2 were mixed with a 10%1 molar excess
of oxygen gas. This mixture was passed through the 33 cm.
reactor held at 280o. Upon completion of the run, 38.9 g. of
material were obtained. This material was distilled to give
the results listed in Table 4.


Cut Weight Boiling Molecular
Number in Gram Range Possible Material Weight

Overhead 20.2 -6o4 COF2, CF3COF 67

1 9.2 -6.o.to -310 CF3COF 118-158

2 3.h -310 to -22o CF3CF2COF 158-166

3 .8 -220 to -9.70 CF300F2COF' 166-181
Flat at -10

Residue 09 -9.60 and up -

The 29.4 g. of COF2 and CF3COF were found to contain
64.4 mole percent COF2. On the basis of reaction (6), 0.13
mole of olefin was consumed in the formation of COF2 and
CF3COF. This left 0.09 mole of COF2 unaccounted for.
Formation of CF3CP2COF accounted for 0.02 mole of olefin and
formation of C30OCF2COP accounted for 0.03 mole. This
accounted for a total of 0.18 mole of reacted olefin. Since
only 0.04 mole of olefin remained unaccounted for it appeared
impossible for the additional 0.09 mole of COF2 to have been
formed by the oxidation of terminal CF2 groups. .Equation
(7) requires only 0.03 mole of olefin to account for the
excess COF2.
The oxidation of CF3CF:CF2 over NaF was felt to be
gratifying since appreciable yields of useful products were
formed. The major process was found to be the production
of CF3COF and COF2. A new compound, CF3OCF2COF, not reported
in the literature, was prepared in approximately 10% yield.

jxeriments Performed Over Nickel Packina

The olefin oxidation was attempted over an inert

nickel surface in order to establish whether or not the
reaction observed over NaF was catalytic or thermal. It was
felt that a nickel surface would not be catalytically active
for the oxidation reaction. The results of these experiments
are summarized in Table 5. The same procedure was observed
in these runs as was used in the NaF experiments. The



actor ol Ratio Approx. Moles of CF3CF=CF2
Rn N. Reactor Mole Ratio Approx.
Run No.Contact Time
Tep. O2/CF3CF-CF2 in Seconds Initial Recovered Used

1~3 I 2500 1.02 145 0.23 0.23 -

145 I 2800 1.00 10 0.32 0.31 -

h II 3100 1.01 1h3 0.29 0.16 0.13

66 II 300o 1.18 109 0.26 0.25

69 II 3300 1.18 95 0.25 0.25

70 II 3000 1.10 100 0.25 0.15 0.10

TABIE 5 Continued

% by Weight % by Weight
% Conversion of COF2 and of Products Rmarks
to Products CFiCOF in Boiling Above
Products C3C CF2

0 -No detectable reaction
took place.

No detectable reaction
0 took place.

Fluoride ion test on pack-
46 100 ing neg. No high boiling
products formed.

Fluoride ion test on pack-
0 ing neg. No detectable
reaction took place.

Fluoride ion test on pack-
97 82 18 ing poe. High boiling
products formed including
3.3% CF30CF2CO.

Fluoride ion test on pack-
ing pos. High boiling
38 products formed, No in-
vestigation of the high
boiling material was made.

33 cm. reactor was packed with 60.5 g. of protruded 0.16 in.
x 0.16 in. nickel column packing. After each run the
packing was examined to ascertain whether or not NiF2 had
formed on the nickel surface.
No reaction was observed in runs 143 I and 145 I.
Comparing these two experiments with runs 136 I and 33 II
(Table 2) performed over NaF indicated that the NaF surface
exhibited a catalytic effect. The reaction over NaF could
not be due to thermal effects since the contact times over
the Ni surface were approximately double those over NaF.
Thermal reaction was believed to be responsible for
the oxidation that took place in run 4 II. Only two products,

COF2 and CF3COF, were formed in this experiment. This
evidence seemed to indicate that the primary reaction
occurring over a Ni surface was:

CCF3CCF2 + 02 = COF2 + CF3COF. (8)

Two experiments were performed, 69 II and 70 II, so
that the mole ratio of these two products could be measured
in order to verify equation (8). Unfortunately, this aim was
not accomplished as NiF2 was formed on the Ni surface in
both of these runs. The measured ratio was approximately
one mole of CF3COF to two moles of COF2. The effect of
NiF2 on the reaction is illustrated by the observation that
run 66 II gave no conversion whereas run 70 II gave 38%
conversion. Both experiments were performed at 3000.

Furthermore, no high-boiling products were observed in

run 4 II, in the absence of NiF2, but high boiling products
were formed in 69 II and 70 II where NiF2 was known to be
A possible explanation of this lies in the fact
that some water vapor was present in run 69 II. The products
obtained in run 69 II contained a good deal of HF. The
receiving trap was etched. This water vapor was introduced

by changing drying tubes in the reactant stream, prior to

run 69 II, and by not flushing the system out with dry
It was felt that the water vapor present in run
69 II hydrolyzed some of the products to form HF which at
this elevated temperature attacked the Ni packing to form
NiP2 which then acted as a catalyst for the oxidation
reaction in run 70 II. This observation was born out by
the fact that the Ni packing gave a positive fluoride ion
test and that high boiling products were obtained in runs
69 II and 70 II. No high boiling products were formed in run
4 II where fluoride ion was absent.

Exerints Performed Over CsF

Five experiments were performed over CsF. These
experiments, tabulated in Table 6, offer convincing proof
that inorganic fluorides catalyze the oxidation of
CF3CF:CF2. For these runs, a 15 cm. reactor packed with
119.9 g. of CsF was used. These experiments were performed


Moles of CFgCF*CF2
Reactor Mole Ratio Contact Time __ of Cp3CF
Run o Temp. 02/CF3CCF2 in Seconds Initl Recoved

59 II 2500 1.10 l3.5 0.17 0.17 0

61 II 2800 1.10 37.0 0.15 0 0.15

63 II 2800 1.10 3.5 0.13 0 0.13

6 II 240o 1.10 40.8 0.12 0.01 0.11

52 III 215 1.1 16.0 0.79 0.33 0.16

TABLE 6 Continued

% by Weight % by Weight
% Conversion of COP2 and of Products r
to Products COF Boiling Above

0 -No detectable reaction.

Reaction mixture too com-
.99 58 l2 plex to separate by

Reaction mixture too com-
99 60 40 plex to separate by

Reaction mixture too caa-
95 56 44 plex to separate by

58 44 56 10% CF3CF2COF.

in the order listed. It is interesting to observe that run
59 II produced no conversion whereas run 64 II produced
approximately 95% conversion despite the fact that each run
was made under essentially equivalent conditions. It was
noted that two runs were made at 2800 where essentially
complete conversion was obtained, and one run, 52 III, was
made at 2150 where 58% conversion was obtained. Thus it

appeared that the sample of CsF used in run 59 II had no
catalytic properties. From this it was concluded that
catalytic properties were introduced into this sample by

first initiating the oxidation reaction at a higher
temperature, namely, 280.

Experiments 61 II, 63 II and 64 II yielded mixtures
of high boiling products that were not separable by
distillation. This was due to the small quantities
initially used and to the poor distillation characteristics
of mixtures of similar fluorocarbon substances. Run 52 III
was performed in an attempt to obtain a sufficient quantity
of this material to permit separation by distillation. This
run produced a more favorable ratio of high-boiling products
to low-boiling products than did any of the other experiments.
This run is discussed in some detail in the following

Experiment 52 III: One-hundred and eighteen grams
(0.79 mole) of CF3CF:CF2 were reacted with 28.7 g. (0.90 mole)
of oxygen over CsF at 2150. The 128 g. of products

obtained were distilled to give the results tabulated in
Table 7.
The overhead material and Cut 1 were scrubbed through
NaOH solution to yield 7.8 g. of unreacted olefin. The
ratio of COF2 to CF3COF was not determined for this material.
Cut 2 contained 41.4 g. of unreacted CF3CP:CF2. This
CF3CF:CF2 combined with the 7.8 g. of CF3CF:CF2 recovered
from Cut 1 gave 49.2 g. of unreacted olefin.
Cut 3 was assumed to be mainly CF3CF2COF with a
trace of CF3COCF3 present as was the case when the oxidation
was performed over NaP. This cut also contained intercut
boiling up to -19.3.
Cut 4 consisted mainly of unidentified intercut.
The distillation curve for this material contained a slight
inflection at -110 to -100. This was interpreted as
indicating that a small amount of CF30CF2COF was present in
this material.
Cut 5 was found to be essentially pure CF3COCF2CF3.
Molecular weight remained constant over the 1.40 distillation
range. This ketone was made in 21% yield assuming that two
molecules of CF3CF=CF2 were required for its formation.
Cut 6 contained 9.9 g. of unidentified material.
This cut showed a small flat at 26.50 to 27.50 where 2.2 g.
of material of molecular weight 261 were collected. The
column hold up, excluding the pot residue, was pumped out of
the column and also collected in this fraction. No



Possible material


Overhead 1.8 -6U to -30.20 COF2, CF3COF up to 150

2 41.L -30.2o to -28.2o CF3CF-CF2 150-157

3 7.4 -28.2 to -19.3 CF3CF2COF, 157-170

10.1 1930 to +00 Intercut,

.1 -19.3 to +1.0 Saome CF0CF2COF 170-216

5 10.6 1.0 to 2.4h CF3COCF2CF3 216

6 9.9 2.j0 to 28.50 -- 216-261

Pot 5.2 up to 600 (CF3)2CFCOCF2CF 329-339


in Grams



identification was made of the material boiling between
26.50 and 27.5.
The pot residue was believed to be chiefly

(CF3)2CFCOCF2CF3. The boiling point of this material was
estimated from the pot temperature to be in the range 500
to 600.
The results obtained by oxidizing CF3CF=CF2 over CsF
were considered to be very significant. Considerably lower
temperature was required and a greater yield of high-boiling
products was obtained. This fluoride was found to be the
most active of all the materials studied. Unfortunately,
the complex nature of the products made it impossible to
compute an acceptable material balance for the products.

The Attempted Oxidation of CF3CF2CF3

Over CaF

The success obtained in the oxidation of CF3CF:CF2

over CsF led to an attempted oxidation of the saturated
fluorocarbon CF3CF2CF3. This experiment was performed in

the same apparatus used for the olefin oxidations. The

CF3CF2CF3 was mixed with an equal volume of oxygen in a
water operated storage bottle. This mixture was passed

through a 15 cm. reactor packed with CsF. The experiment
was started at 4000 and the temperature was slowly raised to
6200 for the last part of the experiment. Contact times
varied from 45 to 35 seconds. Upon completion of the

experiment the CF3CF2CF3 was quantitatively recovered. Thus
it was concluded that this method of oxidation was unsatis-

factory for saturated fluorocarbons.

Determination of the COF2, CF3COF Ratio

This determination was made by reaction the COF2
and CF3COF, obtained during the distillation of the reacting
products, with a NaOH solution and then measuring the CO2
liberated upon acidification. In each of the determinations
listed in Table 8, distillate was collected until the
boiling point of CF3CF=CF2 was reached. This distillate
was combined with the material obtained overhead. These
combined products were reacted with a 20%, carbonate free,
NaOH solution.
The product ampule was equipped with two stopcocks
so that the entire sample could be made to react by
sweeping with nitrogen gas. Any unreacted material was then
swept successively through Drierite and P205 drying tubes
by the nitrogen stream. These products were collected in
another trap, equipped with stopcocks, cooled in liquid air.
This material was then degassed and weighed. In every case
it was found that the unreacted material obtained in this
manner was CF3CF=CF2. The weight of CF3CF=CF2 thus

obtained was deducted from the original weight of products
reacted to give the true weight of COF2 and CF3COF involved.



Run No. 33 II 56 II 62 II 63 II 64 II 69 II 70 11

Reaction Tep. 2800 2800 2800 2800 240 3300 300

Catalyst NaF NaF CoF CsF CaF Ni Ni

Total Weight
COF2, CF3COF 29.4 20.1 16.3 13.7 13.0 35.3 9.8

Weight of Olefin
W t of 0.1 1.6 0.1 0.1 0.1 0.9 3.6

Total Weight
COF2, CF3COF 29.3 18.8 '26.2 13.6 12.9 314. 6.2

Molecular Weight
Molecular Wig .5 4.2 43.8 44.0 45.1 13.7 14.2
of C02

Mole % COF2 65.6 61.7 67.0 66.5 71.4 59.3 67.5

Mole % COF2 64.4 61.3 65.1 66.7 62.9 59.3 61.0

The NaOH solution containing the reaction products

was now slowly acidified with H20S4 while stirring

vigorously. The evolved CO2 was passed successively through
Drierite and P205 drying tubes. Upon completion of acid-
ification the apparatus was flushed with nitrogen gas to
insure that all the evolved C02 was collected in the liquid
air-cooled end trap.
This C02 was then degassed and weighed. From this
weight the amount of COF2 originally present could be
calculated. This value is listed in Table 8 in the column
headed Mole Percentage COP2 Gravimetric. This weighed
quantity of CO2 was then introduced into a calibrated volume
in the vacuum system. The sample was allowed to volatilize
and reach room temperature. The pressure exerted by this
sample in the known volume and room temperature were then
measured. Using the ideal gas relation it was then possible
to calculate the amount of COF2 originally present. Values
obtained in this fashion are listed in Table 8, in the
column headed Mole Percentage COF2 Volumetric.
A molecular weight determination was made on each
sample of C02 measured. These values are listed in Table 8,
in the column labeled Molecular Weight C02.
As a check on this analytical procedure a blank
determination was made. Twenty grams (0.13 mole) of pure
CF3CF=CF2 were subjected to the exact procedure used in
the analysis determinations. It was found that no

measurable quantity of C02 was evolved thus proving that
the NaOH solution was carbonate free.

Examination of Table 8 shows that in each case a
COF2 to CF3COF ratio of approximately two to one was obtained.

Some Conclusions Drawn fro the

Preliminary Work

The work done in this research indicated that

classical oxidation catalysts such as CuO and Hopcalite are

not satisfactory agents for the oxidation of fluorine

containing compounds. The major process in oxidations using

these materials was found to be the production of 002 and

metal fluorides. Useful products were not isolated in
appreciable yields.
Oxidations performed over inorganic fluorides were
much more successful. Inorganic fluorides have an advantage
over oxide type catalyst in that they do not offer a final

resting site for fluoride ions. Possibly the traces of

products found with the oxide catalysts used were due to the
formation of fluorides which then acted as the active


Experiments over NaF gave products that were totally

unexpected. Most of the products found required fluorine

migration to explain their structures. Considering the

surfaces upon which these reactions took place this is not
difficult to understand.

The oxidations over CsF were more successful than

those over NaF from the standpoint of products formed. The

ratio of materials boiling above CFsCF:CF2 to those boiling
below CF3CF:CF2 was appreciably increased using CsF as
contrasted to NaF. These two fluorides show a great
difference in selectivity in the various products formed.
The product CF30CF2COF was formed in good yield over NaF
but was not isolated over CsF. On the other hand, CF3COCF2CF3
was not found when NaF was used but was made in good yield
over CsF. The catalytic nature of this oxidation over NaF
and CsF appeared to be well established. Comparison of the
temperature used and the products obtained over a clean Ni
surface with the results over NaF and CsF substantiated
this fact.

It was felt that these preliminary experiments could
be done under a controlled set of standard conditions to

evaluate the relative efficiencies of the alkali fluorides
as oxidation catalysts. The acidic nature of the majority
of the products formed was used as the basis of analysis
for the comparison experiments described in Chapter VI.



Only one product of any importance, other than C02,

was formed over the oxide catalysts. The identification of

this product is discussed first. The remaining products
discussed were all formed over inorganic fluoride surfaces.
A number of piperazine derivatives of perhalo-acids were
prepared to aid in identification of the products. The
properties of these derivatives are reported in Table 12.

Identification of CF3C(OH)2CF3

When CF3CF:CF2 was oxidized over CuO or Hopcalite,
fine, needle like, low melting crystals were often found in
the wet ice-cooled product trap. When the products contained
in the Dry Ice-cooled and liquid air-cooled traps were
transferred into the vacuum system more of these crystals
were left behind as a residue. In experiment 65 I the
crystals obtained from these two sources were combined in

the vacuum system to give 0.5 g. of a solid that melted at
370 to 390 in a sealed tube under its own vapor pressure.
Two determinations on the vapor showed the molecular weight
to be 172 and 184 at pressures of 15.5 a.m. and 22.2 m.m.

respectively. No significance was attached to this

variation in molecular weight with pressure because of the

+1.0 m.m, error in pressure readings.
This material was divided into a more volatile and a
less volatile fraction in the vacuum system by sublimation.
The less volatile fraction melted at 41.50 to 42.50. This
substance was water soluble and gave a neutral solution that
contained no fluoride ion. A small amount of this material
was used in a sodium fusion. The acidified product gave a
positive fluoride ion test.
Another 0.4 g, of this solid was obtained as a

residue when the volatile products from the reaction were
transferred into the distillation column. This material was
whiter than that obtained from the wet ice-cooled trap. This
substance was water soluble, neutral, and gave a negative
fluoride ion test as before. This material was divided into
two fractions, by sublimation, each fraction having a melting
range of 450 to 46.
An infrared spectrum run on the first fraction
showed no carbonyl band, thus eliminating the possibility of
a ketone type structure. The adsorption bands observed for
this substance in CC14 solution are noted in Table 9.
Karyakin and Nikitin34 tabulated the infrared
spectra of nine hydroperoxides and seven peroxides. They
reported a hydroxyl valency vibration at 3450 ca. -1.
They found that there are no vibrations characteristic of


Peak Percentage Possible
Cas.1 Transmission Interpretation

3550 50 -OH Valency Vibration
1210-1245 5.0 typical -C-F Adsorption
1163 15 -C-O- Stretching
925 47 -C-0-
715 35

-C-O-O-Ci although there are three frequently appearing
frequencies, 860 (-0-0-), 940 and 1200 ca.-1(2C-O-). This
solid was subjected to a modified peroxide test.35 The
negative result of this test indicated the absence of a
peroxide linkage.
The problem of structure was cleared up when another
sample of this unknown was prepared. From run 120 I, 1.7 g.
of this material was obtained. The solid exhibited a vapor
pressure of approximately 24 m.m. at 300. The calculated
molecular weight was in the range 182 to 192 allowing for an
error in pressure measurement of 1.0 a.m. A melting point
determination showed this material to melt in the range
44.50 to 46.00. No change in melting range occurred when
this solid was sublined into two portions. These two
portions were recombined and were repeatedly passed through
a P205 drying tube. This treatment produced a low boiling
liquid of molecular weight 167.

Comparison of the infrared spectrum of this gas with
that of a known sample of hexafluoroacetone,36 (theoretical
molecular weight 166), showed them to be identical. Since
no carbonyl band is present in the infrared spectrum of the
solid and the molecular weight of the solid is different
from that of the gas by about eighteen units it was felt that
the formula CF3C(OH)2CF3 was a correct representation for
the structure of this product.
The formation of perfluoro-ketone hydrates has been
observed before.37-40 Generally the pure hydrates were not
obtained. Hydrate formation has also been observed with the
perfluoro-aldehydes. The aldehyde hydrates are reported to
be isolatable as solids.41,42 Henne39 reported the formation
of a white solid, melting below room temperature, during the
distillation of CF3COCF3. This solid appeared wherever moist
air had access to the distillation system. This solid gave a
semicarbazone derivative melting very near hexafluoroacetone
semicarbazone prepared from an aqueous solution. Morse43
reported a white solid when CF3COCF3 was fractionated on the
vacuum line. This material gave a mass spectrum identical
with that of CF3COCF3. When sealed in a glass tube it was
easily sublimed by applying the heat of the hand. They
considered the substance to be a polymeric form of CF3COCF3
and provisionally assigned to it the structure of a trimer.
In view of the information collected in this
research, it was felt that the solid observed by these

workers was merely the gem-diol form of hexafluoroacetone

In the course of identifying this material a
polarographic reduction was performed on the second fraction
of the material melting at 450 to 460. When 0.1 N
(CH3)4N Br (aqueous) was used as the supporting electrolyte,
the unknown gave a well defined reduction curve with a half
wave potential of -2.02 volts with respect to a saturated
calomel electrode. The nature of the reduction products was

Identification of COF2

This product was found to be the main constituent in
the material that came off overhead, past the Dry Ice-acetone
cooled condenser, during distillation. In a typical
experiment, 132 I, 45.9 g. of material were obtained in this
fashion. A simple trap to trap, one plate distillation was

made and molecular weights were determined on approximately
equal fractions. The results cf these determinations are
listed in Table 10.
A comparison of a known44 infrared spectrum trace of
COF2 with the infrared spectrum on the material of molecular
weight 66.5 showed this substance to be COF2 (theoretical
molecular weight, 66.0). This material was found to be
completely soluble in basic solution. Upon acidification,
a molar quantity of CO2 equivalent to the quantity of
COF2 dissolved was liberated.


Fraction Molecular Possible
Number Weight Material

1 68.3 COF2
2 66.5 COF2
3 106.8 COF2, CF3COF
4 113.0 CF3COF

Identification of CF COF

This material was usually found in the fraction

coming off overhead from the distillation column. Relatively
pure samples could be obtained through the vapor take-off
outlet once the major portion of COF2 had been removed. In

a typical experiment, 33 II, a small flat was obtained at
-59.50 to -59.0. This material had a molecular weight of
117 and was believed to be pure CF3COF (theoretical molecular
weight, 116, B.P. -590). The products coming off overhead
combined with those distilling between -64 and -31 were
reacted with an excess of NaOH solution. The resulting
solution was neutralized with HC1. A considerable quantity
of CO2 was liberated but was not measured. After
neutralization, the mixture of salts was carefully taken to
dryness in a 1100 oven. The dry salts were extracted with
three successively smaller portions of absolute ethyl

alcohol to remove any CF3COONa present. Evaporation of the

solvent gave the sodium salt. This salt was washed into a
small side arm distillation flask. Enough concentrated

H2SO4 was added to make the concentration in the distillation
flask about 50%. This solution was steam distilled and the
trifluoroacetic acid liberated was collected in a small test
tube containing cold propanol-2. This solution was divided
into two portions. A piperazine derivative was prepared
from one portion and a N-phenylpiperazine derivative was
prepared from the-other portion. The derivatives were
recrystallized until a constant melting point was obtained.
A mixed melting point determination was made for each salt
with a sample of the known derivative of trifluoroacetic
acid. In either case no depression of the melting point was

Identification of CF3CF2COF

In a typical experiment 134 I, this material was

contained in the cut distilling from -28.2o to -26.7.
Molecular weights on this cut ranged from 155 to 167. An
infrared spectrum on the center cut of this material agreed
with that reported44 for CF3CF2COF (theoretical molecular
weight, 166, B.P. -28).
A sample of this cut was reacted with a known
quantity of water in the vacuum system. After three hours
reaction time the remaining volatile material that had not

reacted was trapped in a liquid air-cooled trap. This

material was dried through a P205 tube. This product was
weighed and a molecular weight determination showed this

material to be unreacted CF3CF:CF2.

The remaining hydrolyzed material was now treated
with an excess of NaOH in the vacuum system. Over a period
of several hours, a gas was slowly liberated. This gaseous
material is discussed in the next section. The liquid
remaining after NaOH treatment was neutralized with HC1 and
taken to dryness, care being taken to prevent overheating of

the salts. The dry salts thus obtained were extracted with
two portions of absolute ethyl alcohol. The ethyl alcohol

was then taken to dryness and the salt obtained was again

extracted with two small portions of absolute ethyl alcohol.

The solvent was evaporated off to yield a quantity of
sodium salt.

The free acid was liberated from this salt by
acidification followed by steam distillation. A piperazine
derivative was prepared from one portion of the acid and
a N-phenylpiperazine derivative was prepared from another
portion. These derivatives were repeatedly recrystallized
to eliminate the small trace of trifluoroacetic acid
derivative that was known to be a contaminant. A mixed
melting point determination made for each derivative, with
a sample of the known derivative, showed no depression.
This information in conjunction with the boiling point,

molecular weight, and infrared spectrum was conclusive proof

that this reaction product was CF3CF2COF.
The formation of this material in appreciable yield
was considered unusual. It was felt that this material
could possibly arise from the oxidation of CF3CF2CF2H. This
possibility was checked as the above mentioned hydride is
often a contaminant of CF3CF:CF2 when prepared by the
decarboxylation of CF3CF2CF2COONa.33
A sample of the starting material, CF3CF:CF2, was
dried through a P205 tube and then degassed by freezing and
thawing while pumping away any evolved gases. Molecular
weights taken on the first and last fraction were 149.8 and
150.1 respectively, both in good agreement with the
theoretical value of 150.0. This sample was subjected to
infrared analysis. The C-H bond in CF3CP2CF2H causes it
to show adsorption3 in the 2940 to 2990 cm." region. No
adsorption band was found in this region. From the above
evidence it was concluded that the starting material did not
contain any appreciable quantity of hydride. Thus it was
concluded that CF3CF2COF did not arise from the oxidation of

Identification of CF3COCF3

In the previous section a gas was reported liberated
when the aqueous solution of products was treated with
excess NaOH in the vacuum system. This gas was collected

and weighed. The molecular weight was determined to be 70.1.
An infrared spectrum determination showed this material to
be CF3H (theoretical molecular weight, 70.0) which would
arise from the hydrolysis16 of CF3COCF3. It was impossible
to separate this ketone in a pure state since its boiling
point was so close to that of CF3CF2COF. The quantity of
CF3COCF3 produced in this experiment, calculated from the
weight of CF3H, was about 1% of the theoretical maximum.

Identification and Proof of Structure


This new acid fluoride was contained in the
fractions distilling from -10.50 to -9.50. This material
had a molecular weight of 182, fumed in air and had an
unpleasant odor. This substance hydrolyzed in water with
the liberation of considerable heat. A test for fluoride
ion on this aqueous solution was positive.
A special piece of apparatus was constructed so that
accurately weighed samples could be reacted without loss of
material. Duplicate samples were reacted with a known
quantity of standard NaOH solution. The aqueous solutions
were back titrated with standard HC1 solution to give an
equivalent weight of 91. These same solutions were
analyzed for fluoride ion by the thorium nitrate technique.45
Duplicate determinations gave a value of 10.32% hy-
drolyzable fluoride ion. These analytical results suggested

an empirical formula of C3F602 (calculated equivalent
weight, 91; hydrolysable fluoride ion, 10.44%). Chromato-

graphic analysis indicated that this unknown was a single
compound. A peroxide test performed on this material was
negative and thus eliminated the possibility of a hypo-

fluorite group. Cady46 reported that this group reacts with
KI solution to liberate iodine explosively.

An infrared spectrum on this material was determined
with care being taken to maintain anhydrous conditions.

The principle adsorption bands and their possible interpre-
tations are reported in Table 11.
From the spectra trace it was apparent that a COF
group was present. This point was confirmed in the neutral
equivalent and fluoride ion determinations.
A sample of this unknown was treated with excess
NaOH in a vacuum system. No gaseous products were evolved


Wave No. Percentage Possible
Ca." Transmission Interpretation
1905 1.5 C=0 stretching,
acid fluoride
1360-1080 0.0 typical C-F adsorption
908 3.0
832 43.7

indicating the absence of ketone groups. The NaOH solution
was neutralized with H2SO4 and taken to dryness in a 1100
oven. The resulting solids were extracted with absolute
ethyl alcohol to yield a quantity of the sodium salt of the
unknown. This sodium salt was found to be deliquescent.
This salt was acidified with 50% aqueous H2S04, in a small
side arm flask, and the free acid was steam distilled over
between 980 and 108. The distillate was found to be free
of fluoride ion. This aqueous solution had an odor resembl-
ing perfluorobutyric acid. One-half of this solution was
neutralized with N-phenylpiperazine to give a water
insoluble oil, as a lower layer, which after separation and
drying at 1100 failed to solidify. This material could not
be recrystallized from the usual solvents. The other half
of the solution was neutralized with anhydrous piperazine to
yield an insoluble solid. This material was recrystallized
from ethyl alcohol and water to give a white solid melting at
191.00 to 191.50. These results did not agree with the
properties of any of the known fluorocarbon acid piperazine
salts. The results of an elemental analysis on this
compound, which showed it to be 2C3F602, C4H10N2, are
reported in Table 12.
The two possible isomers of C3F602 were CF3CF20COF
and CF30CF2COF. The first of these was expected to be
unstable to basic hydrolysis as it would be a derivative
of carbonic acid. The second isomer should be stable.

To differentiate between these two possibilities
the sodium salt of a sample of the acid was prepared. This
salt was decarboxylated according to the procedure of Henne47
by refluxing for three hours in the presence of ethylene
glycol. The reaction products were collected in liquid air.
These products were passed through an Ascarite drying tube
to remove CO2. Molecular weights were determined on the
remaining gas. The first fraction over gave a molecular
weight of 134.5 and the middle fraction gave a value of
135.5. An infrared spectrum on this material indicated that
it was CF30CF2H (calculated molecular weight, 136). Although
no information about this ether was found in the literature,
its spectrum was quite similar to the spectrum of CF30CF3.48
This was considered to be proof that the structure of this
unknown was CF3OCF2COF.

Identification of CF3COCF2CF3

This material was contained in the fractions
boiling from 0.5o to 2.00. In a typical experiment, 52 III,
distillate was collected between 1.00 to 2.4. A molecular
weight determination on the center cut of this material
gave 216 (calculated molecular weight, 216). This material
was water soluble, but went into solution slowly. An
infrared spectrum on this material indicated the presence
of a ketone carbonyl group. Holub reported49 that

CF3COCF2CF3 boiled at 0.00, was water soluble and gave a

semicarbazone melting at 1370 to 1380, with decomposition.
A semicarbazone derivative prepared on an aqueous solution
of the material obtained in this work melted with
decomposition at 1350 to 1360. The yield of derivative was
very low.

Tentative Identification of


This product was obtained as a pot residue. The
boiling point was estimated from the pot temperature to be in
the range 500 to 600. A sample of this material was washed
with water to remove traces of acidic materials. The
sample was then dried through a P205 tube. Vapor density
measurements showed the.molecular weight to be in the range
319 to 329 (calculated molecular weight for C6F120, 316).
The material was washed with a dilute solution of NaOH. A
gas was observed to be slowly liberated. This gas was
dried through a P205 tube and collected. Upon completion
of gas evolution a small amount of insoluble material
remained unreacted. This unreacted material was discarded.
A molecular weight determination on the evolved gas gave a
value of 171. Infrared spectra determined on the first and
last fractions of this material were identical. Comparison
with known spectra50 showed this material to be (CF3)2CFH
(calculated molecular weight, 170).

This hydride could have arisen from the preferential
hydrolysis of (CF3)2CFCOCF2CF3. The symmetrical ketone
(CF3)2CFCOCF(CF3)2 would produce only one hydride but this
possibility seemed to be eliminated by its high molecular
weight (366) and boiling point. Hauptschein40 reported
that C3F7COC3F7, an isomer of (CF3)2CFCOCF(CF3)2, boiled at
75. Haszeldine37 found that C3F7COC2F5, an isomer of
(CF3)2CFCOCF2CF3, was water insoluble and boiled at 52.
This was in fair agreement with the properties of the
compound found in this work. Unfortunately, the sodium salt
of the other hydrolysed portion of this molecule was
discarded. The preparation of a piperazine derivative of
this material would have solved the problem.

Piperazine and N-Phenylpiperazine Derivatives
of Some Perhalo-Acids

In the course of this research, it was desirable to
find a method for the identification of perfluoro-acids.
Pollard and others51-53 demonstrated that the salts of
piperazine were very useful in the identification of organic
The piperazine and N-phenylpiperazine salts of the
more readily available perhalo-acids were prepared. These
salts were found to be readily purified and their melting
points were in a convenient range.
These derivatives and their properties are

summarized in Table 12. Melting points were determined in



Analyses*, Percentages
Perhalo-Acid M.P. C
Corr. Calculated Found
C H F C0 C H F Cl

Piperazinium Salts



cF3 (cF2) cooH










Above 250

Above 250


























-- 2

-- 2



-- 30.1

-- 29.5

-- 28.3

-- 26.6

-- 26.5

.6.2 23.9

7.1 22.8

- 33.2

- 27.1















- 26.41

- 27.1

31.7 -

12.8 -


TABLE 12 Continued

Analyses*, Percentage*
Perhalo-Acid ** oC
lM.Calculated Found
C H F C1 C H F Cl
N-Phenylpiperazinium Salts

CF3COOH 151-154 52.2 5.18 20.6 52.3 5.18 20.8 -
CF3CF2COORH 14-114.;5 7.8 1.64 29.1 17.7 1.52 29.1 -
CF3(CF2)2COOH 124-126 44.7 1.02 35.4 -- 4.3 1.00 35.6 -
CF(CF2)COOH 122-121 10.3 3.18 44.0 -- 1.0 3.10 43.8 -
CF3(CF2)6COOH 125.5-128 37.5 2.62 49.5 37.6 2.62 50.1 -
Cl(CF2CFC1)2CF2COOB 133-131 36.6 2.88 20.2 36.8 3.00 20.1
C1(CF2CFC1)3CF2COOH 141-112 33.7 2.36 22.1 33.6 2.71 21.9

*Analyses by Clark Microanalytical Laboratory and Schwarzkopf Microanalytical Laboratory.

sealed tubes in a brass block which was preheated to within
50 of the melting point. The experimental procedure used in
the preparation of the piperazinium salts is described in
the following paragraphs.
To about 1 cc. of the acid, contained in a small
beaker, was added 10 cc. of propanol-2. The beaker was
placed in an ice bath, and anhydrous piperazine was slowly
added with stirring until the resulting mixture was just
basic to litmus paper. Another 10 cc. portion of propanol-2
was added and the slurry was filtered while cold.
The crude product was recrystallized by adding 10 cc.
of 95% ethanol and adding enough water dropwise to the
boiling mixture to cause the salt to dissolve. The solution
was cooled in ice until crystallization was complete. This
procedure was repeated four times. On the last recrystal-
lization the solution was filtered with suction while hot.
The material from the last two recrystallizations showed no

change in melting point.
The piperazinium salts of C1(CF2CFC1)2CF2COOH,

C1(CF2CFC1)3COOH and perfluoroglutaric acid were recrystal-
lized from boiling water. They were extremely insoluble in
all solvents, and were only successfully recrystallized in
minute amounts from hot, filtered aqueous solutions.
The piperasinium salt of perfluorooctanoic acid was
recrystallized by filtering a hot saturated solution of the

salt in n-butanol. This salt was extremely insoluble in
all solvents.
The experimental procedure used in the preparation
of the N-phenylpiperazinium salts is described in the
following paragraphs.
To about 4 cc. of the acid, contained in a small
beaker, was added 20 cc. of propanol-2. The beaker was
placed in an ice bath. N-phenylpiperazine was slowly added
to the acid solution with stirring until the resulting
slurry was just basic to litmus paper. The mixture was
heated until solution of the solid was complete. The
crystals which separated when the flask was cooled were
filtered and recrystallized four times from propanol-2. In
the last recrystallization the solution was filtered while
hot. The material obtained in the last two recrystalliza-
tions showed no change in melting point. The N-phenylpiper-
azinium salts of Cl(CF2CFC1)2CP2COOH and C1(CF2CFC1)3COOH
were recrystallized from absolute ethanol.



General Procedure Used in the Comparison

Sufficient oxygen was included in the feed mixture
to insure essentially complete oxidation if the conditions
of temperature and contact time warranted it. That sufficient
oxygen was present was deduced from the major reactions that

were found

the alkali

to occur in





2CF3CF=CF2 +

2CF3CF=CF2 +

the preliminary experiments involving

These reactions were:

02 = CF3COF + COF2 (8)

1%02 = 3COF2 (9)

i02 CF3CF2COF (10)

i02 CF3COCF3 (11)

02 CF30CF2COF (12)
li02 CF3CF2COCF3 + 2COF2 (13)

i02 (CF3)2CFCOCF2CF3 (14)

This reaction series accounted for about 95% of the
products formed. The maximum stoichiometric ratio of oxygen
to olefin occurred in reaction (9). Since this reaction
only accounted for 13% to 16% of the total products it was

not necessary to use enough oxygen to satisfy this equation
to insure complete oxidation. The feed material ratio of
oxygen to olefin was set at 1.135. This arbitrary ratio
was used for all of the comparison experiments. This
mixture was prepared by adding a given quantity of olefin
to the calibrated 27 1. storage bottle. The calculated
amount of oxygen was then added, from a storage cylinder, in
the same manner. A small quantity of water was always left
in the storage bottle so that the gases could be intimately
mixed by vigorous agitation of this water. This standard
mixture was then metered into the reactor at the desired
flow rate.

Prior to starting a series of experiments, the
reactor and catalyst were evacuated at 4000 for two hours
by pumping through a liquid air trap. Upon completion of
this treatment dry nitrogen was admitted to bring the pressure
in the reactor up to atmospheric.
Reaction products were collected in a liquid air
cooled glass trap, equipped with stopcocks. Upon completion
of a run the reaction system was flushed with dry nitrogen
for 15 min. to insure that all products were collected in
the end trap. This trap was now connected to the vacuum
line and any unreacted oxygen was removed by a process of
pumping, thawing, refreezing and then repumping until all
noncondensables were removed. The total weight of products
was then recorded.

This trap was connected to the analytical system
where the products were passed through a 20% NaOH solution.
Upon completion of this scrubbing the reaction train was

swept with nitrogen for 20 min. This procedure swept all
unreacted substances successively through Drierite and P205
drying tubes into the end trap which was immersed in liquid

air. This trap was degassed and the contents weighed. The
material recovered in this fashion was assumed to tbe only

CF3CFPCF2. From this value and the calculated amount of
olefin used initially it was possible to compute a percentage
conversion value.
In each series of runs involving the alkali
fluorides the oxidation reaction was initiated at a tempera-
ture considerably higher than that required for reaction
once the catalyst had been activated. It was found that
the best reproducibility was obtained when the runs over the

alkali fluoride salts were performed in order of decreasing


The weighing technique used in these runs introduced
some error into the percentage conversion values. A new
triple beam balance, reading to 0.1 g., was used in all

weighing. The traps containing products were cooled in
liquid air to about one inch above the product level. The
remainder of the trap was dried with a soft cloth and any
ice adhering to the cold portion was removed before weighing.
This same technique was used to obtain empty weights on the

traps. All weights were recorded to the closest 0.1 g.
As pointed out in Chapter II, an individual reactor
was built for each salt. The dimensions of these reactors
were held constant in order to minimize geometry effects.
Contact times were held as constant as was feasable in an
attempt to eliminate as many process variables as was
It was found that special care had to be taken in
loading the reactors containing KF, RbF and CsF because of
the extremely deliquescent nature of these materials. CsF
was especially bad in this respect. A special copper funnel
was made to fit snugly inside the inlet tube of the reactor
and yet not reduce the size of the opening appreciably.

Prior to loading one of the salts into the reactor it was

dried on a hot plate at 3500 in a platinum vessel. A heat

lamp was placed directly over this vessel to help keep the
salt dry. The hot, dry salt was then transferred by spatula
into the copper funnel, attached to the dry reactor, which
was heated by the heat lamp. If this procedure was not
observed it was found that these deliquescent salts would
sinter into one hard lump in the reactor upon heating even
though they appeared dry when loaded into the tube.
The deliquescent nature of these salts made it
difficult to maintain a constant flow rate because end plugs
would form where the reactant gases were admitted to the
reactor. These plugs increased the pressure drop across

the reactor and therefore interefered with the Flowrator
calibration. These difficulties were minimized after some

experience had been obtained in running the equipment.
Information pertaining to the individual catalysts and
reactors used in the comparison experiments is found in
Table 13.
In Figures 2-9 each run number has a point number
associated with it. Point numbers are used to clarify the

text associated with each figure. Point numbers indicate

the sequence in which the experiments were performed.

Comparison Experiments Using LiF

Twelve experiments were performed over a new sample
of LiP. These experiments are tabulated in Table 14.

Considerable difficulty was encountered with this salt in
maintaining activity. Fig. 2 indicates that experiments
92 II and 93 II (represented by points 1 and 2) were
performed before the catalyst was activated. Maximum

activity was assumed to have been obtained after completion
of experiment 94 II. These experiments were performed on

the same day. Points 4, 5, 6 and 7 were determined on the

following day and show no irregularities. The catalyst was

then allowed to stand at 2000 in an atmosphere of dry

nitrogen for three days. Points 8 and 9 were then determined.

These two experiments clearly show that the catalyst had
lost activity. The catalyst was then reactivated at



Reactor Reactor
Catalyst Catalyst Catalyst Total Wt. Catalyst Vol. Vol.
Form Density Catalyst Vol. Empty Packed
g./cc. g. cc. cc. cc.




































Ni Protruded 8.9 2.6 4.8 53.9 49.1

CaF2 Powder 3.2 59.2 18.6 52.1 33.5

BaF2 Powder 4.8 68.0 14.1 52.1 38.0






Dry Vol. Calc.
Contact Feed Wt. of Total Wt. Wt. of % Conv.
Run No. Temp. Time Mixture Olefin Products Olefin to
Sec. Used Used Formed Recovered Products
cc. g. g. g.

92 II 324.50 24.8 2875 8.1 8.0 7.4 9

93 II 345 25.0 2875 8.1 8.0 7.8 4
94 II 380o 25.3 3833 10.8 12.9 0.6 94
95 II 3630 22.0 3823 10.8 13.4 0.6 94
96 II 3470 25.7 3823 10.8 13.4 0.7 94

97 II 3290 26.5 3823 10.8 13.1 0.5 95
98 II 307.50 24.4 3823 10.8 13.1 0.8 93
99 II 282.5 25.8 1911 5.4 5.2 5.1 6
100 II 2920 32.1 1911 5.4 5.5 5.2 4
101 II 2920 27.9 2875 8.1 9.4 2.6 68
102 II 2880 26.6 1911 5.4 5.9 3.0 44

104 II 28L4 24.5 1911 5.4 5.6 5.3 2


280 300 320 340 360 380


Run No. Point No.

Run No. Point No.


Fig. 2. Comparison Experiments Using LiF

0 LLO 1- I

350 by passing two liters of the reaction mixture through
the reactor. After reactivation, points 10, 11 and 12 were
Molecular weight checks on the unreacted olefin
showed that essentially no impurity was present.

Comparison Experiments Using NaP

Two series of experiments were performed using a
reactor containing new NaP pellets. The first set of runs
was performed at the usual contact time. The second set of
runs was done at approximately one-half the usual contact
time values. These experiments, performed over a period of
three days, are listed in Table 15.

Contrary to the behavior noted with LiF, no diffi-

culty was encountered in activating this fluoride.

The curves obtained in these two sets of experiments

are shown in Fig. 3. Reducing the contact time shifted the
percentage conversion curve to the right as was expected.
The small shift observed in these two sets of runs indicated
that minor variations in contact time, within a given set of
runs, did not greatly affect the shape of the curve.

Comparison Experiments Using KF

The six experiments performed over KF are tabulated

in Table 16. The last four runs which determined the
conversion curve, shown in Fig. 4, were done in one day. No


Dry Vol. Calc.
Contact Feed Wt. of Total Wt. Wt. of % CoT.
Run No. Temp. TimL Mixture Olefin Products Olefin to
Sec. Used Used Formed Recovered Products
cc. g. g. g.

Experiments at Standard Contact Time

106 II 3W10 22.7 1911 5*. 6.5 o.4 93
108 II 3270 18.8 2867 8.1 8.3 7.7 5
109 II 337.50 19.9 1911 5.4 5.6 2.0 63
110 II 326.50 20.8 2007 5.8 5.8 5.9 o
111 II 3580 18.7 3823 10.8 13.1 0.k 96

Experiments at One-half Standard Contact Time

112 II 360W 9.3 2875 8.1 10.2 1.2 85

113 II 3530 9.5 2875 8.1 10.0 1.7 79
113 11 3390 9.3 2395 6.8 7.3 5.7 16
115 II 3820 10.4 32148 9.2 11.2 0.6 93
116 11 3440 9.8 2875 8.1 8.8 1.7 1.2

340 360


Run No. Point No.
106 II 1
108 II 2
109 II 3
110 II 4
111 II 5

Run No.
112 II
113 II
114 II
115 II
116 II

Point No.

Two Sets of Comparison Experiments Using NaF





1 I I I


7 0



/ Standard C.T.

0) One-Half
) 8 Standard C.T.

4 I I







- I

Fig. 3.


Dry Vol. Calc.
Contact Feed Wt. of Total Wt. Wt. of % Conv.
Ran No. Temp. Time Mixture Olefin Products Olefin to
See. Used Used Formed Recovered Products
cc. g. g. g.

85 II 2780 20.4 3823 10.8 10.9 10.8 0

.86 I 2960 19.9 2867 8.1 7.3 10

87 II 316 19.7 3823 10.8 12.8 1.5 86

88 II 328 19.2 1911 5.4 6.6 0.5 91

89 II 297 20.1 1911 5.4 6.0 2.7 50

90 II 284 20.3 2867 8.1 8.3 7.8 4



O 3


50- O -



01 02
o I I--I I I I I
260 280 300 320 340 360


Run No. Point No.

85 II 1
86 II 2
87 II 3
88 II 4
89 II 5
90 II 6

Fig. 4. Comparison Experiments Using KF

previous experiments had been done with this KF containing
reactor. The first experiment, run 85 II, produced no
conversion. The temperature was raised to 3800 and reaction
was initiated. Two liters of mixture were allowed to pass
through the reactor at this temperature. The reactor was
then swept out with nitrogen and run 86 II was performed.
Maximum activity had not yet been attained. The catalyst was
successfully reactivated by again raising the temperature to
3800 and allowing four liters of reactant mixture to flow

through the reactor. Four more points were then determined.
Molecular weight checks on the recovered olefin
showed essentially no impurity to be present.

Comparison Experiments Using RbP

This reactor contained new, unused RbF. The
activation behavior of RbF was similar to that of LiP. RbF
was not active until reaction was initiated and maintained
at a higher temperature than was required for reaction over
the activated catalyst.
Activation required initiation of the reaction at
3800. Approximately four liters of ihe feed mixture were
passed through the reactor before activity was obtained.
Fig. 5 shows the result of insufficient activation of the
catalyst on the activity. From this figure it can be seen
that points l, 2 and 3 were determined before the catalyst

was at maximum activity.





T, C

Point No.

Run No.


Point No.

Fig. 5. Comparison Experiments Using RbF




- I I I I I I

6 5




09 1
- 010



Run No.




Dry Vol. Calc.
Contact Feed Wt. of Total Wt. Wt. of % Cnv.
Run No. Temp. Time Mixture Olefin Products 01efin to
Sec Used Used Formed Recovered Products
cc. g. g. g.

120 II 2920 24.2 2895 8.2 5.8 29
121 II 319.50 24.5 3860 11.0 12.6 1.6 85
122 II 3100 24.4 3860 11.0 13.4 1.0 91
123 II 2910 22.8 3860 11.0 12.8 0.8 93
124 II 2800 25.2 3860 11.0 12.9 0.8 93
125 II 2690 24.6 3860 11.0 13.3 1.3 88
126 II 260.50 24.8 38W2 10.9 12.4 3.9 64

127 II 252 23.2 3812 10.9 11.8 5.9 6
128 II 238.50 24.2 3842 10.9 11.6 7.2 34
129 II 2330 25.0 2407 6.8 7.4 5.0 26
131 II 2270 23.9 382 10.9 11.0 10.2 6

Eleven experiments, reported in Table 17, were done
with RbF in order of decreasing temperature. These runs were

performed over a period of five days. No loss in activity
was observed due to the catalyst sitting idle. During
periods when no experiments were being performed, the
catalyst was maintained at 2000 in an atmosphere of nitrogen.
Some relatively non-volatile products were formed
over RbF in the temperature range 2400 to 2700. Molecular
weight checks on the recovered starting material indicated

that some impurities were present.

Comparison Experiments Using CsF

The reactor containing CsF had previously been

used in four preliminary experiments. These experiments were
runs 59 II, 61 II, 63 II and 64 II. The results of these

experiments showed that this catalyst was activated. Nine
runs were made over a period of five days with CsF. Fig. 6

shows that no loss of activity occurred with this catalyst
during this time. When no reaction was being performed,
the catalyst was maintained at 2000 in an atmosphere of
nitrogen. The conditions used and the results obtained in
these experiments are listed in Table 18.
When the material in the product traps from runs

77 II, 78 II and 79 II was being analyzed it was noted that
some of the products were not swept through the analysis

system even by nitrogen. No attempt was made to force this


T, C

Point No.

Fig. 6. Comparison Experiments Using CsF




2 O








Run No.
75 II
76 II
77 II
79 II
80 II
81 II
82 II
83 II


Dry Vol. Calc.
Contact Feed Wt. of Total Wt. Wt. of % Cony.
Run No. Temp. Time Mixture Olefin Products Olefin to
Sec. Used Used Formed Recovered Products
cc. g. g. g.

75 II. 2650 28.0 3832 10.8 11.8 1.4 87

76 II 2L47 30.0 3823 10.8 12.6 1.7 84
77 II 2350 28.9 3823 10.8 12.5 2.4 78
79 II 2250 24.8 3823 10.8 12.0 3.6 67
80 II 203.50 25.6 3823 10.8 11.9 4.6 58
81 II 1850 26.5 3812 10.7 11.0 8.1 24
82 II 1750 27.4 3812 10.7 11.0 9.8 8

83 II 1600 28.6 3823 10.8 11.1 11.1 0

non-volatile substance through the NaOH solution as this

material was obviously not olefin. These non-volatile
products were probably the same as those found in the

preliminary experiments and would contribute some error to

the percentage conversion values. Although no measurement
of quantities involved was made, there appeared to be more
non-volatile material formed over CsF than RbF. Assuming
ketones to be present, as was the case in the preliminary
experiments, their hydrolysis in the analysis system would
give rise to unreactive hydrides. These hydrides along with
all other unreactive products that entered the analysis
system were swept into the end trap and were weighed as
unconverted CF3CF:CF2. Molecular weight checks showed that
some material other than CF3CF:CF2 was present in the
material recovered from the runs over CsF.

Comparison Experiments Using Nickel

The nickel reactors used in the preceding experiments
appeared to be exceedingly resistant to attack from the
reaction products at temperatures up to 380., In view of
this a nickel surface was tried in the belief that it would
not be active for the oxidation reaction. It was felt that
this experiment would establish an upper temperature limit
where thermal reaction would set in.
Four runs were made with a reactor that initially
contained clean nickel packing. These runs are tabulated

in Table 19. From Fig. 7 it is seen that nickel packing
became active after two runs weremade. This activity was
caused by the nickel surface becoming coated with NiP2. A
fluoride ion test performed after experiment 36 II was
positive although there was very. little change in the
physical appearance of the nickel packing.

Comparison Experiments Using CaF2

Four runs were performed with the reactor containing

CaP2. These runs are tabulated in Table 19, and the
conversion curve is shown in Fig. 8. Run 50 II seemed to
indicate that this surface also became active after use.
However, no further experiments were performed to verify

Comparison Experiments Using BaF2

Seven runs were performed over BaF2. These runs are
tabulated in Table 19. These experiments fell into two
groups. Fig. 9 shows that experiments 39 III, 40 III and
41 III determine one curve and experiments 42 III, 43 III and
44 III determine another similar curve. The first group of
runs (39 III, 40 III and 41 III) were performed on the same
day. The second group (42 III, 43 III and 44 III) were
performed one day later. These two curves merge at high
percentage conversion values. It is possible that the
variation between these two sets of data is within the

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs