Gas-phase NMR studies of the thermolysis of perfluorooctanoic acid

Article abstract of DOI:10.1016/j.jfluchem.2005.08.016

The thermolysis of perfluorooctanoic acid (CF₃(CF ₂)₆COOH) has been studied kinetically by high-temperature gas-phase NMR in both sodium borosilicate glass and quartz ampoules. In both cases, 1-H-perfluoroheptane is the major product, but the decomposition is considerably slower in quartz. The decomposition in borosilicate ampoules at 307°C is greatly accelerated by the presence of crushed borosilicate glass and the reaction appears to be completely heterogeneous. Perfluoro-1-heptene is produced as a minor product and time-of-flight secondary ion mass spectrometry (ToF-SIMS) examination of the inner surface of the borosilicate glass ampoule after reaction reveals the presence of NaF. The thermolysis in quartz ampoules was studied by ex situ heating in the temperature range 355-385°C and produced moderate amounts of perfluoro-1-heptene and SiF₄ in addition to 1-H-perfluoroheptane. Thermolysis in the presence of added crushed quartz accelerates the decomposition, as does increasing the concentration of perfluorooctanoic acid or carrying out the thermolysis in the presence of water. The thermolysis of perfluorooctanoic acid thus proceeds at widely different rates depending on concentration and on the physical and chemical environment. By contrast, the pyrolysis of ammonium perfluorooctanoate is more facile by orders of magnitude and proceeds by first-order kinetics at essentially the same rates in both quartz and borosilicate ampoules with an estimated half-life of 2 s at 307°C.

Full text of DOI:10.1016/j.jfluchem.2005.08.016

Journal of Fluorine Chemistry 126 (2005) 1510–1516
www.elsevier.com/locate/fluor
Gas-phase NMR studies of the thermolysis of
perfluorooctanoic acid§
1
Paul J. Krusic , Alexander A. Marchione, D. Christopher Roe
*
DuPont Corporate Center for Analytical Sciences, Central Research and Development,
Experimental Station, Wilmington, DE 19880, USA
Received 5 July 2005; received in revised form 29 August 2005; accepted 30 August 2005
Available online 2 November 2005
Abstract
The thermolysis of perfluorooctanoic acid (CF3(CF2)6COOH) has been studied kinetically by high-temperature gas-phase NMR in both sodium
borosilicate glass and quartz ampoules. In both cases, 1-H-perfluoroheptane is the major product, but the decomposition is considerably slower in
quartz. The decomposition in borosilicate ampoules at 307 8C is greatly accelerated by the presence of crushed borosilicate glass and the reaction
appears to be completely heterogeneous. Perfluoro-1-heptene is produced as a minor product and time-of-flight secondary ion mass spectrometry
(ToF-SIMS) examination of the inner surface of the borosilicate glass ampoule after reaction reveals the presence of NaF. The thermolysis in quartz
ampoules was studied by ex situ heating in the temperature range 355–385 8C and produced moderate amounts of perfluoro-1-heptene and SiF4 in
addition to 1-H-perfluoroheptane. Thermolysis in the presence of added crushed quartz accelerates the decomposition, as does increasing the
concentration of perfluorooctanoic acid or carrying out the thermolysis in the presence of water. The thermolysis of perfluorooctanoic acid thus
proceeds at widely different rates depending on concentration and on the physical and chemical environment. By contrast, the pyrolysis of
ammonium perfluorooctanoate is more facile by orders of magnitude and proceeds by first-order kinetics at essentially the same rates in both quartz
and borosilicate ampoules with an estimated half-life of 2 s at 307 8C.
# 2005 Elsevier B.V. All rights reserved.
Keywords: Pyrolysis of perfluorooctanoic acid; High-temperature gas-phase NMR; Kinetic NMR; Ammonium perfluorooctanoate
1. Introduction
The mechanism for thermolysis of trifluoroacetic acid (TFA,
the lowest PFCA homolog) in the gas phase has been examined
previously and these studies [6–9] have shown that the
decomposition reaction is both complex and subject to wall
effects (surface assistance). By analogy with the rates of
decarboxylation of TFA and its anion under hydrothermal
conditions [10], it seems likely that undissociated perfluor-
ooctanoic acid is thermally more stable than any of its salts.
Due to the low pKa of 1.5 for perfluorooctanoic acid [11], the
anion is the dominant form expected to be present in solution.
Conditions that suffice for the thermal decomposition of the
free acid are, therefore, expected to exceed the conditions
required for the thermolysis of any of its salts.
Ammonium perfluorooctanoate (APFO) is a surfactant
commonly used as a processing aid in the production of
fluoropolymers and fluoroelastomers [1–3]. As a class of
surfactants, perfluorocarboxylates (PFCAs) and perfluorosulfo-
nates are remarkably stable and it is this property, that makes
them useful in many applications [1]. It has been asserted that
PFCAs have no known environmental degradation pathway [4].
It has recently been shown that APFO does not survive the
elevated temperatures (350–400 8C) specified for fluoropolymer
processing [5]. We were, therefore, interested to examine the
thermal stability of the related PFCA, perfluorooctanoic acid.
We have examined the kinetics of thermal decomposition of
perfluorooctanoic acid using high-temperature gas-phase
NMR. Gas-phase NMR has previously been used to study
the chemical kinetics of a variety of processes [5,12–14]. While
the acronym PFOA has been used in the literature either to
represent the perfluorooctanoate anion or to refer ambiguously
to either the acid or the salt, in this report the abbreviation
§
DuPont Contribution Number 8624.
* Corresponding author. Tel.: +302 695 3593; fax: +302 695 9799.
E-mail address: chris.roe@usa.dupont.com (D.C. Roe).
1
Retired.
0022-1139/$ – see front matter # 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.jfluchem.2005.08.016

P.J. Krusic et al. / Journal of Fluorine Chemistry 126 (2005) 1510–1516
1511
PFOA for perfluorooctanoic acid (CF3(CF2)6COOH) strictly
refers to the free acid.
2. Results and discussion
Previous gas-phase NMR studies in our laboratory utilized
sodium borosilicate glass ampoules (e.g., Ref. [5]). In view of
the role of surface processes in the thermolysis of TFA [6–9],
alternative materials were considered for the present study.
Quartz ampoules flame-dried under vacuum provide the most
inert reactor surface compatible with our gas-phase NMR
method. A typical sample in a sealed quartz ampoule
consisted of 32 mg of PFOA (77 mmol, m.p. ꢀ55 8C, b.p.
ꢀ200 8C) and 23 mmol of C2F6 as internal mass and chemical
shift standard. Assuming ideal gas behavior and an internal
volume of 4.0 mL, the internal pressure at 307 8C is close to
1 atm and the molar concentration of PFOA is approximately
0.02 M. The gas-phase 19F NMR spectrum at 307 8C shows
six resolved CF2 fluorine resonances and a resonance for the
CF3 fluorines.
Fig. 1. Kinetics of gas-phase PFOA (*) decomposition at 370 8C in a quartz
ampoule and appearance of 1-H-perfluoroheptane (&), perfluoro-1-heptene
(^) and SiF4 (~). The linear best fit of the mass balance (dashed line)
represents the sum of the NMR intensities of PFOA, 1-H-perfluoroheptane
and perfluoro-1-heptene.
In situ NMR kinetics at 307 8C for 65 h in this relatively
inert reactor led to approximately 4% conversion of PFOA,
formation of approximately 2% of 1-H-perfluoroheptane [5]
and a trace of SiF4. Proton NMR detection at the end of the
reaction likewise revealed the formation of 1-H-perfluorohep-
tane (d 5.76, 1H, t, J = 52 Hz) as well as the formation of water
(d 0.60 in the gas phase). The concomitant formation of CO2
(and possibly CO), containing neither proton nor fluorine, falls
outside our methodology and is assumed; similarly, non-gas-
phase reaction products are not observed.
tripled. Essentially, all PFOA (96 mg) was decomposed in
about 2 h instead of about 9 h (32 mg), indicating that
molecular associations play an accelerating role at higher
concentrations. Similarly, decomposition was markedly accel-
erated at 365 8C by the presence of two equivalents of added
water and was complete in about 45 min.
Decomposition of carboxylic acids in thegas phase is prone to
heterogeneous effects and surface assistance [16]. In order to
evaluate the role of quartz in the observed kinetics, the in situ
thermolysis of PFOA at 307 8C was repeated in the presence of
approximately 320 mg of flame-dried crushed ampoule quartz as
a means of increasing the surface-to-volume ratio. The reaction
was substantially accelerated by the presence of the crushed
quartz (Fig. 2), demonstrating that heterogeneous reactions are
likely also to play a role in the quartz ampoules used for the ex
situ kinetic runs. Approximately, 30% conversion of PFOA was
obtainedafter65 h, incontrasttoonly 4% conversionobserved in
the absence of crushed quartz. 1-H-Perfluoroheptane and SiF4
are the most visible products, growing steadily as the reaction
proceeds and there are only traces of perfluoro-1-heptene (<1%)
at the end of the kinetic run.
PFOA decomposition at higher temperatures was investi-
gated by ex situ heating in quartz ampoules since the
temperatures for convenient kinetics exceed the in situ
capabilities of our NMR probe. Heating of the ampoules
was carried out in a furnace at 355, 370 and 385 8C for
appropriately chosen times, followed by quenching to room
temperature. The extent of reaction was then monitored by 19
F
NMR at 243 8C. Thermal decomposition of PFOA at 355–
385 8C gives two organofluorine products, 1-H-perfluorohep-
tane as the major one and perfluoro-1-heptene [15] as a minor
one. SiF4 also grows in as a product as the decomposition
reaction proceeds.
The top line in Fig. 2 represents the time dependence of the
normalized intensity of the CF3 fluorine resonance
(ꢁ81.2 ppm) that incorporates the CF3 signals of all species
in the vapor phase having CF3-terminated perfluoroalkyl
chains, as these CF3 fluorines all have the same chemical
The time evolution of the major reaction components in
these ex situ experiments was fitted by first-order exponentials
as shown in Fig. 1 for the kinetic run at 370 8C. The integrated
NMR intensity of the mass standard is invariant and is used to
normalize the NMR intensities of the reaction components. The
sum of the concentrations of the principal organic components
(PFOA, 1-H-perfluoroheptane and perfluoro-1-heptene)
decreases slightly during the kinetic run (vide infra). Similar
first-order fits were obtained for the kinetics at 355 and 385 8C.
The pseudo first-order rate constants for loss of PFOA,
formation of 1-H-perfluoroheptane and formation of perfluoro-
1-heptene are presented in Table 1. The estimated half-life for
PFOA at 385 8C is 26 min under these conditions. Evidence for
lack of truly first-order behavior was obtained by an analogous
kinetic run at 370 8C in which the amount of starting PFOAwas
Table 1
Apparent first-order rate constants for gas-phase decomposition of PFOA in
quartz ampoules
Temperature (8C)
k ꢂ 103 (minꢁ1
PFOA
)
CF3(CF2)5CF2H
CF3(CF2)4CF CF2
355
370
385
5.0 ꢃ 0.7a
9.2 ꢃ 0.7
7.0 ꢃ 1.8
11.1 ꢃ 1.2
23.0 ꢃ 2.1
1.8 ꢃ 0.9
4.3 ꢃ 0.9
8.9 ꢃ 4.5
26.4 ꢃ 3.1
a
Errors represent one standard deviation.

1512
P.J. Krusic et al. / Journal of Fluorine Chemistry 126 (2005) 1510–1516
(1)
(2)
Fig. 2. Kinetics of gas-phase PFOA (&) decomposition at 307 8C in a quartz
ampoule containing approximately 320 mg of crushed quartz. The appearance
of 1-H-perfluoroheptane (*), SiF4 (+) and the constant intensity of the C2F6
standard (^) are indicated. The dashed line is the sum of the NMR-derived
concentrations of PFOA and 1-H-perfluoroheptane. The upper line (~) is the
normalized NMR intensity of the terminal –CF3 fluorine resonance common to
all long-chain species in the vapor phase.
: CFRF þ HꢁF ! HCF2RF
(3)
(4)
shift. By the end of the 65 h reaction time, the intensity of this
signal drops by about 7%, indicating a loss of such CF3-
containing molecules from the gas phase. The NMR intensities
of PFOA and 1-H-perfluoroheptane, on the other hand, add up
to the dashed line in Fig. 2. There are thus unaccounted
molecules with overlapping CF3 resonances in thevapor phase.
Indeed, several minor peaks grow steadily during the course of
the kinetics, but have not been positively assigned: d ppm
(relative strength) ꢁ67.0 (vw), ꢁ70.4 (w), ꢁ137.3 (vw),
ꢁ153.5 (vw), ꢁ157.2 (vw), ꢁ163.1 (w), ꢁ163.8 (w). In the
spectral region of –CF2 resonances there are additional
growing peaks obscured by severe overlap. The chemical shifts
ꢁ67.0 and ꢁ70.4 are very close to the chemical shifts for cis
and trans perfluoro-2-butene [12] and the remaining chemical
shifts are appropriate for cis and trans vinylic fluorines [17].
Isomerized internal perfluoroheptenes, particularly 2-perfluor-
oheptene, are, therefore, candidates for the unaccounted
species.
The results obtained with added crushed quartz unmistakably
implicate heterogeneous accelerating processes on quartz and
make it impossible to distinguish between heterogeneous and
homogeneous reactions in the thermal decomposition of PFOA
under the current conditions. Accordingly, the temperature
dependence of the gas-phase NMR kinetics presented in
Table 1 cannot be interpreted mechanistically. The apparent
activation parameters obtained from this kinetic data for PFOA
decomposition are Ea = 197 kJ molꢁ1 and log (A/sꢁ1) = 12.2.
The rate of gas-phase decomposition of PFOA is, therefore,
expected to vary depending upon the nature of the surface
available.
A surface reaction occurring in unison with the formation of
1-H-perfluoroheptane and 1-perfluoroheptene is the formation
of SiF4 on the quartz (silica) surface (Fig. 1) and is anticipated
to involve HF. The formation of commensurate amounts of
water (SiO2 + 4HF ! 2H2O + SiF4) and the probable isomer-
ization of perfluoroheptene are both consistent with the
involvement of HF, as required by Eqs. (2) and (4) or by
their heterogeneous analogs.
It is appropriate to consider the possible initiating
unimolecular processes (Eqs. (1) and (2)) analogous to those
considered in the pyrolysis of TFA (RF = F) [6–9,18].
Decomposition via
a
four-membered transition state
(Eq. (1)) was excluded both experimentally and theoretically
and may likewise be excluded for PFOA (RF = CF3(CF2)4CF2).
Elimination of HF via a five-membered transition state (Eq. (2))
with the formation of difluorocarbene (RF = F) was accepted as
the initial step for TFA decomposition and formation of CF3H,
analogous to 1-H-perfluoroheptane, was ascribed to the
insertion of difluorocarbene into the H–F bond (Eq. (3),
RF = F). In the case of PFOA, Eq. (2) would lead to
fluoro(perfluorohexyl) carbene (:CF(CF2(CF2)4CF3). In con-
trast to TFA, longer-chain perfluorocarboxylic acids have the
potential to decarboxylate with loss of HF via a six-membered
transition state to produce the corresponding terminal
perfluoroalkene (Eq. (4)).
Since the fluorine mass balance in the vapor phase slowly
decreases with time (Fig. 2), the presence of involatile fluorine-
containing species was sought by time-of-flight secondary ion
mass spectrometry (ToF-SIMS) mapping of the inner wall of
the reacted quartz ampoule. In contrast to an unreacted
ampoule, which appeared relatively uniform, the reacted
ampoule was decorated with domains of high ion-yield that
were approximately 30 mm in diameter. Mass spectra extracted
from these domains revealed several progressions of mass
peaks from approximately 1000–1500 Da differing in mass
by 50 Da appropriate for CF2 fragments. The formation of
high-molecular weight species is consistent with the involve-

P.J. Krusic et al. / Journal of Fluorine Chemistry 126 (2005) 1510–1516
1513
ment of carbenoid intermediates that can lead to polymers
[19,20].
thermolysis experiment at 307 8C was repeated in the presence
of approximately 370 mg of this crushed glass, the decom-
position was faster than any observed so far under our
conditions at this temperature. Approximately, 90% of the
PFOA was consumed in the first 4.5 h and after a total of 8 h,
99% of the PFOA was consumed (Fig. 4) compared with only
30% conversion after 65 h at the same temperature under
analogous conditions in quartz. 1-H-Perfluoroheptane and
perfluoro-1-heptene were formed as the major products during
the first 4.5 h and no significant further reactions took place
(attesting the stability of the products under these conditions).
Only traces of SiF4 were observed and the relative yield of the
olefin was substantially greater than in the absence of crushed
glass. Proton NMR examination after the reaction revealed the
characteristic –CF2H proton resonance of 1-H-perfluorohep-
tane and a peak for water at 0.6 ppm. Comparison of the
fluorine and proton spectra indicated that water was formed in
an amount equivalent to the olefin.
The above evidence for heterogeneous reactions on silica
suggested that PFOA decomposition could behave differently
in ampoules made of sodium borosilicate glass. The kinetics of
thermolysis of PFOA in such an ampoule at 307 8C is presented
in Fig. 3. The concentration of olefin is approximately 5% that
of 1-H-perfluoroheptane and the mass balance indicates that
formation of 1-H-perfluoroheptane and perfluoro-1-heptene
completely accounts for the loss of PFOA. Considering that in a
quartz ampoule, only 4% conversion of PFOA was observed
after 65 h at the same temperature, it is clear from Fig. 3 that
sodium borosilicate glass strongly accelerates the decomposi-
tion of PFOA relative to quartz. In contrast to the analogous
decomposition in quartz (Fig. 1), SiF4 was either absent or
present in trace amounts, in similar experiments in borosilicate
glass.
The time dependence in Fig. 3 is unusual in that it appears to
be bimodal and suggests that surface sites that are initially
capable of catalyzing the decomposition reaction are being
deactivated. A simple model starting with a first-order equation
in which the rate constant is allowed to be time-dependent and
to decrease in time in proportion with the number of converted
molecules leads to decays that can be represented by a sum of
two first-order exponentials.
The thermolysis of PFOA in the presence of crushed glass
corresponds to the regime where active surface sites are in large
excess relative to reactant molecules. Under these conditions,
the decay of PFOA and the formation of both 1-H-
perfluoroheptane and perfluoro-1-heptene, follow pseudo
first-order kinetics with very similar rate constants of
9.0 ꢂ 10ꢁ3, 9.8 ꢂ 10ꢁ3 and 7.9 ꢂ 10ꢁ3 minꢁ1, respectively
(in agreement with the theory for parallel first-order reactions
[21]).
In the limit of a much larger number of active sites relative to
reactant molecules, pseudo first-order is obtained. In the
opposite limit, when there are more reactant molecules than
active sites, the rate tends to zero when all sites are deactivated
and the reaction greatly slows down. The data presented in
Fig. 3 can be fitted reasonably well by this model, but the fitting
parameters have no general significance since they refer only to
this specific set of experimental conditions.
The decompositions of PFOA in quartz and in borosilicate
glass differ in another respect. While SiF4 is a product of
reaction growing in unison with the two major organofluorine
products in quartz, it is essentially absent in borosilicate glass.
To produce the olefin, it is necessary to lose a fluorine for which
there is no evidence in the gas phase given the mass balance
(Fig. 4). It may be considered that the fluorines removed to
make perfluoro-1-heptene react with and stay on the
borosilicate glass. One key difference between silica and
borosilicate glass is the presence of sodium ions in the latter. It
is, therefore, probable that the sodium salt of PFOA is formed
by reaction of the acid with the surface (forming water) and
Further evidence for the involvement of surface sites in the
decomposition of PFOA in borosilicate glass was obtained by
carrying out the kinetics in the presence of crushed borosilicate
glass in order to increase the surface-to-volume ratio. When the
Fig. 3. Kinetics of gas-phase PFOA (*) decomposition at 307 8C in a sodium
borosilicate glass ampoule and formation of 1-H-perfluoroheptane (&) and
perfluoro-1-heptene (^). The mass balance (dashed line) was estimated from
the sum of the NMR-derived organofluorine concentrations.
Fig. 4. Kinetics of gas-phase PFOA (*) decomposition at 307 8C in a sodium
borosilicate glass ampoule in the presence of crushed glass and formation of 1-
H-perfluoroheptane (&) and perfluoro-1-heptene (^).

1514
P.J. Krusic et al. / Journal of Fluorine Chemistry 126 (2005) 1510–1516
then easily decomposes pyrolytically to the olefin (Eq. (5)) as is
well known for such salts [22,23]:
the half-life at this temperature is approximately 1.3 h.
Increasing the concentration of PFOA, as well as carrying
out the decomposition in the presence of water, accelerates the
reaction, indicating that molecular associations provide more
facile decomposition pathways. The thermal decomposition of
perfluorooctanoic acid thus proceeds at widely different rates
depending on concentration and on the physical and chemical
environment. By contrast, the pyrolysis of the ammonium salt,
APFO, is more facile by orders of magnitude and proceeds by
first-order kinetics at essentially the same rates in both quartz
and borosilicate ampoules. From the derived activation
parameters, a half-life of 2 s at 307 8C is estimated for APFO.
It is anticipated that most other salts of PFOA will thermally
decompose at temperatures substantially lower than those
required for the free acid [22,23].
CF3ðCF2Þ4CF2CF2COONa
! CF3ðCF2Þ4CF¼CF2 þ CO2 þ NaF
(5)
ToF-SIMS mapping of a segment of the inner surface of the
borosilicate ampoule supports the notion of this surface reac-
tion. This technique revealed more or less continuous fluorine-
containing domains, which were absent in an unreacted
ampoule. The mass spectral data associated with these areas
showed evidence for formation of sodium fluoride arising from
the fluorine removed to form the fluoroolefin. In total, these
results indicate that the thermal decomposition of PFOA in the
presence of sufficient borosilicate glass is a heterogeneous
process.
4. Experimental
Previously, the in situ thermolysis of solid ammonium
perfluorooctanoate was studied by monitoring the evolution of
1-H-perfluoroheptane (Eq. (6)) in sodium borosilicate glass
ampoules using gas-phase 19F NMR [5]. This study estimated
Ea = 154 ꢃ 11 kJ molꢁ1 and log (A/sꢁ1) = 13.6 ꢃ 1.2 (error
estimates represent 95% confidence limits). It was, therefore, of
interest to examine APFO thermolysis in flame-dried quartz
ampoules in order to ascertain whether the APFO decomposi-
tion process was also sensitive to the type of reactor surface.
Since details of the gas-phase NMR apparatus and
methodology have been described previously [5,13,24], only
a brief outline is presented here. Gas-phase NMR ampoules
were made from 6.5 cm sections of 10 mm o.d. NMR tubes
(borosilicate glass or quartz) fitted with a 5 mm o.d. extension
which served to attach the ampoule to a vacuum system for
sample loading or to center the sealed ampoule in the
thermostated region of the NMR probe (Fig. 5). Micromolar
quantities of PFOA were weighed into the ampoule and a
reference gas (C2F6) was transferred in known micromolar
CF3ðCF2Þ4CF2CF2COONH4
! CF3ðCF2Þ4CF2CF2H þ CO2 þ NH3
(6)
The decomposition of APFO in quartz was studied in the same
temperature range as employed previously (196–234 8C) using
the same amounts of starting material and 79 mmol C2F6
internal mass standard. Although the observed rate constants
in quartz were slightly smaller than found previously, the
associated activation parameters (Ea = 150 kJ molꢁ1 and
log (A/sꢁ1) = 13.0) are well within the error limits estimated
previously [5]. Modest amount of perfluorooctanamide,
CF3(CF2)6CONH2, were also observed. The yield of this minor
product was found to decrease monotonically as the tempera-
ture was increased and extrapolation indicates that the yield of
this minor product would be near zero at 270 8C.
The above observations indicate that the kinetics of
decomposition of APFO is relatively well-behaved, in contrast
to the results observed for PFOA. The activation parameters for
the decomposition of APFO in quartz ampoules may be used to
estimate a half-life of 2 s at 307 8C.
3. Conclusions
PFOA in an isolated environment is thermally quite stable up
to 300 8C, but at higher temperatures is susceptible to
decomposition at increased rates via heterogeneous assistance.
The half-life of PFOA at 307 8C in the presence of crushed
quartz under our conditions may be estimated as 5 days,
whereas in the presence of crushed sodium borosilicate glass,
Fig. 5. Diagram of the gas-phase NMR ampoule attached to a vacuum system
before being sealed off with a propane torch (left). Center: the sealed ampoule
attached by means of heat-shrink tubing to a matching 5 mm o.d. glass stub of
the holding-tube, which rests within a ceramic head. The ampoule assembly is
lowered into the magnet using a string. The amount of sample material is chosen
so that the internal pressure after complete vaporization will not exceed a safe
limit (<1.5 atm).

P.J. Krusic et al. / Journal of Fluorine Chemistry 126 (2005) 1510–1516
1515
amounts prior to sealing the ampoule with a torch; the internal
volume after sealing was approximately 4.0 mL. The added gas
served not only as a chemical shift reference but also as a mass
reference. The starting number of micromoles of PFOA was
obtained from the weight and was compared with the
micromoles of the C2F6 mass standard (as obtained from the
19F NMR intensity) in order to define the starting point for the
kinetics. The integrated NMR intensity of the mass standard
was invariant and was used to normalize the NMR intensities of
the reaction components. Ampoules were obtained from New
Era Enterprises Inc. (Vineland, NJ) and were either kept in a
vacuum oven at 110–120 8C or were flame-dried on a vacuum
line (P < 1 mTorr) before use.
The 19F NMR spectra were obtained using a 400 MHz
Oxford magnet (19F at 376.312 MHz) with a Varian Unity Inova
console. The high-temperature 10 mm probe and variable
temperature controller had a nominal upper temperature rating
of 400 8C and were purchased from Nalorac Corporation (now
Varian Inc.). It was found that in situ kinetic studies lasting
more than a few hours were limited to sample temperatures of
307 8C or less in order to avoid exceeding the upper
temperature limit of the shim stack (110 8C). 19F NMR spectra
were acquired using a small flip-angle pulse (ꢄ 358) in order to
achieve relatively uniform power distribution over the
appropriate spectral region. 19F chemical shifts in the gas
phase were measured relative to C2F6 at ꢁ87.2 ppm on the
CFCI3 (F11) scale. Since 19F spin-rotation relaxation is very
efficient in the gas phase, 250 ms recycle delays were
appropriate and permitted rapid signal averaging (e.g., 180
scans in approximately 1 min). Temperatures were calibrated
using a thin thermocouple positioned in the center of a dummy
ampoule identical to those used for the actual kinetic runs
except for a small hole at the end of the 5 mm stem to permit
insertion of the thermocouple.
shorter than the heating intervals. After a fixed period of time in
the furnace, the ampoule was removed and quickly cooled to
room temperature by covering the ampoule with sand to quench
the pyrolysis. The ampoule was then placed in the NMR probe
at 243 8C, a temperature above the boiling point of PFOA but at
which decomposition was very slow, and the 19F NMR
spectrum acquired as described above.
Time-of-flight secondary ion mass spectrometry data were
acquired with an lon-ToF GmbH Model IV ion-reflector type
time-of-flight instrument (Muenster, Germany). Au+ primary
ions were used and a pulsed electron flood gun was used for
charge compensation. Spatially-resolved ToF-SIMS spectral
data was acquired in 256 pixel ꢂ 256 pixel arrays spanning
500 mm ꢂ 500 mm areas from the inside surfaces of ampoules.
Positively- and negatively-charged secondary ions were
collected in separate experiments. Analysis was carried out
on the inner surface of both reacted and unreacted ampoules.
Linear PFOA (97% purity) was obtained from Oakwood
Products Inc. and was dried in a desiccator and kept in a
nitrogen glove box. 1-H-Perfluoroheptane (1-H-pentadeca-
fluoroheptane) was purchased from Lancaster Synthesis Inc.
and hexafluoroethane (F116) was prepared in our laboratories.
Tetramethylsilane (TMS) and perfluorooctanamide were
purchased from Aldrich and used without further purification.
Perfluoro-1-heptene (>95% purity) was purchased from
SynQuest Laboratories Inc.
Acknowledgments
We are grateful to A.J. Vega (University of Delaware) for
assistance with the numerical integration required to model the
surface site deactivation that leads to fitting of the bimodal
kinetics in Fig. 3. We gratefully acknowledge K.G. Lloyd for
the ToF-SIMS analytical characterization and R.L. Waterland
and B.E. Smart for useful discussions.
The gas-phase 19F NMR spectrum of PFOA at 243 8C
consisted of six CF2 fluorine resonances in the region from
ꢁ116.4 to ꢁ123.9 ppm and a resonance for the CF3 fluorines
(ꢁ81.2 ppm). 1-H-Perfluoroheptane [5] and perfluoro-1-hep-
tene [15] were recognized by resonances that were distinct from
those of PFOA (e.g., terminal –CF2H and perfluorovinyl
References
[1] E. Kissa, Fluorinated Surfactants and Repellants, second ed., Marcel
Dekker, New York, NY, 2001.
fluorines,
respectively).
Perfluorooctanamide
[2] M.M. Schultz, D.F. Barofsky, J.A. Field, Environ. Eng. Sci. 20 (2003)
487–501.
(CF3(CF2)6CONH2) was identified by its H and 19F NMR
spectra which were identical to those of a commercial sample
obtained under conditions similar to those used for the kinetic
studies. The CF3 resonances of the latter three compounds
overlapped that of PFOA. SiF4 was observed at ꢁ169.2 ppm.
Kinetic studies which were performed at temperatures
exceeding 307 8C were conducted by placing the ampoule for
specified time intervals in a furnace consisting of a cylindrical
copper body with a 12 mm bore and holes for thermocouple
placements surrounded by heater wire coils embedded in
ceramic insulation. The temperature controller was from
Electronic Control Systems Inc. (Poway, CA). Furnace
temperatures were checked with a digital thermometer and
agreed within 1 8C up to 400 8C with the temperature read by
the controller. Because of the small heat capacity of the
ampoules, thermal equilibrium was reached in a time much
1
[3] M. Pabon, J.M. Corpart, J. Fluorine Chem. 114 (2002) 149–156.
[4] D.A. Ellis, J.W. Martin, A.O. De Silva, S.A. Mabury, M.D. Hurley, M.P.
Sulbaek Andersen, T. Wallington, J. Environ. Sci. Technol. 38 (2004)
3316–3321.
[5] P.J. Krusic, D.C. Roe, Anal. Chem. 76 (2004) 3800–3803.
[6] P.G. Blake, H.J. Pritchard, Chem. Soc. B (1967) 282–286.
[7] A. Ashworth, P.G. Harrison, J. Chem. Soc. Faraday Trans. 89 (1993)
2409–2412.
[8] D.M. Jollie, P.G. Harrison, J. Chem. Soc. Perkin Trans. 2 (1997) 1571–
1575.
[9] L.M. Reynard, D.J. Donaldson, J. Phys. Chem. A 106 (2002) 8651–8657.
[10] A.J. Belsky, P.G. Maiella, T.B. Brill, J. Phys. Chem. A 103 (1999) 4253–
4260.
[11] Y. Moroi, H. Yano, O. Shibata, T. Yonemitsu, Bull. Chem. Soc. Jpn. 74
(2001) 667–672.
[12] P.M. Kating, P.J. Krusic, D.C. Roe, B.E. Smart, J. Am. Chem. Soc. 118
(1996) 10000–10001.
[13] P.J. Krusic, D.C. Roe, B.E. Smart, Israel J. Chem. 39 (1999) 117–123.

1516
P.J. Krusic et al. / Journal of Fluorine Chemistry 126 (2005) 1510–1516
[14] A.B. Shtarov, P.J. Krusic, B.E. Smart, W.R. Dolbier Jr., J. Am. Chem. Soc.
123 (2001) 9956–9962.
[20] K.K.S. Lau, K.K. Gleason, B.L. Trout, J. Chem. Phys. 113 (2000) 4103–
4108.
[15] A. Battais, B. Boutevin, P. Moreau, J. Fluorine Chem. 12 (1978) 481–496.
[16] W.H. Richardson, H.E. O’Neal, in: C.H. Bamford, C.F.H. Tipper (Eds.),
Comprehensive Chemical Kinetics, vol. 5, Elsevier, New York, 1972, p. 447.
[17] S. Berger, S. Braun, H.-O. Kalinowski, NMR Spectroscopy of the Non-
Metallic Elements, John Wiley & Sons, New York, 1997, p. 445.
[18] J.S. Francisco, J. Chem. Soc. Faraday Trans. 88 (1992) 3521–3525.
[19] S.J. Limb, C.B. Labelle, K.K. Gleason, D.J. Edell, E.F. Gleason, Appl.
Phys. Lett. 68 (1996) 2810–2812.
[21] Z.G. Szabo, in: C.H. Bamford, C.F.H. Tipper (Eds.), Comprehensive
Chemical Kinetics, vol. 2, Elsevier, New York, 1969, , p. 1.
[22] D. Lines, H. Sutcliffe, J. Fluorine Chem. 25 (1984) 505–512.
[23] J.D. LaZerte, L.J. Hals, T.S. Reid, G.H. Smith, J. Am. Chem. Soc. 75
(1953) 4525–4528.
[24] P.J. Krusic, A.A. Marchione, F. Davidson, M.A. Kaiser, C.-P.C. Kao, R.E.
Richardson, M. Botelho, R.L. Waterland, R.C. Buck, J. Phys. Chem. A
109 (2005) 6232–6241.