The effect of oxygen in the photocatalytic oxidation pathways of perfluorooctanoic acid

Article abstract of DOI:10.1016/j.jfluchem.2015.06.019

The influence of oxygen in the photocatalytic oxidation of perfluorooctanoic acid (PFOA) promoted by a commercial nano-sized titanium dioxide was studied by testing the reaction in different conditions: static air, oxygen flux, nitrogen flux and pre-saturated nitrogen flux. The reaction was monitored by Total Organic Carbon (TOC) analysis and Ionic Chromatography (IC). Shorter chain perfluorocarboxylic acids (PFCAs; C_n, n = 1-7) intermediate degradation products were quantitatively determined by High-Performance Liquid Chromatography combined with Mass Spectrometry (HPLC-MS) analysis. The presence of shorter chain PFCAs in solution was also monitored by ¹⁹F NMR. The experimental findings are in agreement with two major oxidative pathways: C_n → C_n-1 photo-redox and β-scissions routes mediated by COF₂ elimination. Depending on the experimental conditions, the mutually operating mechanisms could be unbalanced up to the complete predominance of one pathway over the other. In particular, the existence of the β-scissions route with COF₂ elimination was corroborated by the isolation and characterization of carbonyl difluoride, a predicted fluorinated decomposition by-product.

Full text of DOI:10.1016/j.jfluchem.2015.06.019

Journal of Fluorine Chemistry 179 (2015) 159–168
Contents lists available at ScienceDirect
Journal of Fluorine Chemistry
journal homepage: www.elsevier.com/locate/fluor
The effect of oxygen in the photocatalytic oxidation pathways
of perfluorooctanoic acid
a,b
a
a
c
Maurizio Sansotera , Federico Persico , Valentina Rizzi , Walter Panzeri ,
b,d
b,d
a,b
a,b,
Carlo Pirola , Claudia L. Bianchi , Andrea Mele , Walter Navarrini
*
a
Dipartimento di Chimica, Materiali e Ingegneria Chimica ‘‘Giulio Natta’’, Politecnico di Milano, Via Mancinelli 7, 20131 Milano, Italy
Consorzio Interuniversitario Nazionale per la Scienza e Tecnologia dei Materiali (INSTM), Via G. Giusti 9, 50121 Firenze, Italy
C.N.R. – Consiglio Nazionale delle Ricerche, Istituto di Chimica del Riconoscimento Molecolare, Sezione ‘‘U.O.S. Milano Politecnico’’, Via Mancinelli 7,
0131 Milano, Italy
b
c
2
d
Dipartimento di Chimica, Universit a` degli Studi di Milano, Via Golgi 19, 20133 Milano, Italy
A R T I C L E I N F O
A B S T R A C T
Article history:
The influence of oxygen in the photocatalytic oxidation of perfluorooctanoic acid (PFOA) promoted by a
commercial nano-sized titanium dioxide was studied by testing the reaction in different conditions:
static air, oxygen flux, nitrogen flux and pre-saturated nitrogen flux. The reaction was monitored by Total
Organic Carbon (TOC) analysis and Ionic Chromatography (IC). Shorter chain perfluorocarboxylic acids
Received 8 May 2015
Received in revised form 15 June 2015
Accepted 20 June 2015
Available online 25 June 2015
n
(PFCAs; C , n = 1–7) intermediate degradation products were quantitatively determined by High-
Performance Liquid Chromatography combined with Mass Spectrometry (HPLC–MS) analysis. The
Keywords:
19
presence of shorter chain PFCAs in solution was also monitored by F NMR. The experimental findings
PFOA
are in agreement with two major oxidative pathways: C
mediated by COF elimination. Depending on the experimental conditions, the mutually operating
mechanisms could be unbalanced up to the complete predominance of one pathway over the other. In
particular, the existence of the -scissions route with COF elimination was corroborated by the isolation
n
! Cnꢀ1 photo-redox and
b-scissions routes
Oxidation
Titanium dioxide
Photocatalysis
Carbonyl difluoride
2
b
2
and characterization of carbonyl difluoride, a predicted fluorinated decomposition by-product.
ß 2015 Elsevier B.V. All rights reserved.
1
. Introduction
PFOA and its salts are exogenous very stable perfluorinated
PFOA Stewardship Program by US-EPA and by eight major
companies in 2006 [1]. As a direct consequence of this program,
the fluoropolymer industry has been forced to develop new
environmentally friendly surfactants suitable for emulsion poly-
merization. In addition, from that moment on, many important
research studies on PFOA substitutes and related materials
appeared in the specialized literature [11–17]. Furthermore, due
to recent findings on the extreme difficulty to decompose PFOA by
using standard methodologies, new and effective degradation
techniques like electrochemical [18], modified Fenton reagent
[19], sonochemical [20], plasma [21], microwave [22], photochem-
ical [23,24] and photocatalytic methods [25–29] have been studied
and published. In addition extensive technological reviews are also
available in the literature [30–32].
surfactants, utilized till now for the preparation of the majority
of fluoropolymers that are worldwide employed in thousands of
everyday life essential applications, such as manufacturing,
aerospace, automotive, electronics, semiconductors and textile
[
1]. PFOA UV stability and high surface-active effects are due to its
completely perfluorinated structure [2–4]. However, perfluoro-
carboxylic acids (PFCAs) are nowadays source of great concern due
to their proved persistence in the global abiotic and biotic
environment, including food and humans [5–7]. These compounds,
in fact, are among the most widely diffused fluorinated surfactants
into waste streams [8–10].
The 2015 is the deadline for the complete phase out of
perfluorooctanoic acid; this phase out plane was launched in the
In this study, we analyzed the influence of oxygen in PFOA
photocatalytic oxidation induced by UV-activated TiO
ductors, in particular TiO , are characterized by an electronic band
structure in which electrons from the valence band (V ) are
promoted to the conductive band (C ) with the simultaneous
generation of positively charged holes (h ) in the V
absorption of a photon flux of energy equal or higher to the
2
. Semicon-
2
b
*
Corresponding author at: Dipartimento di Chimica, Materiali e Ingegneria
b
Chimica ‘‘Giulio Natta’’, Politecnico di Milano, Via Mancinelli 7, 20131 Milano, Italy.
Tel.: +39 02 2399 3029; fax: +39 02 2399 3180.
+
b
by the
E-mail address: walter.navarrini@polimi.it (W. Navarrini).
http://dx.doi.org/10.1016/j.jfluchem.2015.06.019
0022-1139/ß 2015 Elsevier B.V. All rights reserved.

1
60
M. Sansotera et al. / Journal of Fluorine Chemistry 179 (2015) 159–168
bandgap energy. The electron–hole pairs can recombine in a few
nanoseconds, or they can be trapped in surface states where they
can react with donor or acceptor species adsorbed on the
photocatalyst surface [33]. Interface redox reactions involving
both excited electrons and photogenerated holes must compete
effectively with the recombination processes of the electron–hole
pairs [34,35]. In the presence of water and oxygen, hydroxyl
into the visible region [48]. Water was purified by using an Elga
Option 3 deionizer and it was used to prepare PFOA solutions for
the different kinetic tests. Milli-Q water was employed for ion
chromatography. HPLC–MS analyses were carried out by using as
1
an eluting phase a mixture of methanol (CHROMASOLV , for HPLC,
1
ꢃ99.9% – from Sigma–Aldrich ) and 2 mM aqueous ammonium
2
acetate solution. The evaluation of COF formation during PFOA
ꢁ
ꢀꢁ
radicals OH and superoxide ions O
2
are generated [36]. These
abatement was specifically monitored by degrading a solution of
1
intermediates are strong oxidizing species able to mineralize
organic compounds [37,38]. However, an extensive debate on the
oxidative pathway of PFOA exists nowadays, because it has been
PFOA in Galden HT-170 (from Solvay Specialty Polymers), a PFPE-
based solvent with boiling point of 170 8C and formula as follows:
CF
.2. Photocatalysis
The photocatalytic apparatus was a 1 L glass stirred reactor
3 2 3 p 2 n 3
O(CF CF(CF )O) (CF O) CF (AMW = 760).
ꢁ
demonstrated that OH radicals generated by Fenton reagent are
2
not very effective in the PFOA mineralization [30,19] and that
IV
ꢁ+
Ti OH can take an active part in the mineralization reaction
[
27,39,40].
1
equipped with an iron halogenide UV lamp (500 W, Jelosil
HG500) emitting light at wavelengths of 315–400 nm and able to
Despite the numerous studies on PFOA oxidation, at the
moment a complete rationalization of the decomposition path-
ways is not available. The interpretation of experimental findings,
as well as a priori calculations reported in the literature, are often in
apparent contradiction [19,26,30] and provide an incomplete
interpretation of the PFOA oxidation mechanism [25] or, in
some cases, the experimental findings are complementary
2
irradiate the reactor with a specific power of 75 W/m . The UV
lamp was placed beside the reactor, which was cooled with water
at a temperature of 30.0 ꢄ 0.5 8C [27]. Titanium dioxide was
introduced in the reactor at the beginning of each test (0.66 g/L)
[
27]. The variation of the surfactant concentration in solution was
monitored by Total Organic Carbon (TOC) analysis and Ionic
Chromatography [27]. The PFOA initial concentration ([PFOA]
mM) was maintained lower than its CMC (7.8 mM [47]) in order to
avoid the formation of emulsions that would reduce the TiO
[
18,21]. The intent of this work is to give our contribution to
0
=
the understanding of this intriguing reaction. In particular, we
focused on the influence of oxygen in the photocatalytic oxidation
4
2
-
2
of PFOA induced by UV-activated TiO and we obtained that,
promoted photodegradation rates [27,28]. Moreover, the PFOA initial
concentration was high enough to allow the detection of the
degradation intermediates, even at very low concentrations. Each
kinetic test was repeated three times in order to evaluate the error
extent and realized by collecting samples (10 mL) of the reaction
mixture at predetermined reaction times. Samples were centrifuged
and filtered through a 0.45 mm polycarbonate membrane in order to
differently from static conditions, a continuous oxygen feeding
enhanced the decomposition of PFOA till its mineralization. On the
contrary, PFOA photooxidation was hindered in a nitrogen-
saturated reaction environment.
The use of excess oxygen directed the PFOA decomposition
through a reaction pathway involving the formation of oxygen-
centered perfluorinated radicals as major intermediates. As
already hypothesized, the perfluorinated oxyradicals preferential-
2
separate the TiO powder from the solution.
Photocatalytic process could be commonly described in terms
of a modified Langmuir–Hinshelwood (L-H) model, which has
been successfully used for heterogeneous photocatalytic degra-
dation by determining the relationship between the apparent
first-order rate constant and the initial content of the organic
substrate [49,50]:
ly followed a
b-scission route by releasing carbonyl difluoride,
COF , as a specific by-product, which can be hardly isolated due to
2
its rapid hydrolysis in aqueous media [41]. Carbonyl difluoride is
an important intermediate for the industrial synthesis of key
fluorinated monomers and it is essential in the preparation of
fluoroplastics and perfluororubbers [42–44]. In this work, we
2
firstly report that the intermediate COF can be isolated in pure
dC
r
S
k K C
form and in good selectivity by performing the photocatalytic
oxidation of PFOA in a suitable perfluorinated aprotic solvent,
instead of water. Moreover, in these conditions the catalyst
deactivation due to fluoride ions is mostly inhibited [27]. As often
happens, a more complete understanding of a reaction mechanism
gives more options in the utility of the reaction under study
r ¼ ꢀ dt
¼
¼ ka p p
S 0
C
C
(1)
1 þ K
In Eq. (1), r is the reaction rate, C is the pollutant concentration
in solution, C is the initial organic content, k is the reaction rate
constant, K is the adsorption rate constant, t is the time and kapp
the apparent first-order rate constant. In the original L-H model the
rate constant k and the adsorption constant K are independent of
light intensity and K should not vary with the light intensity
0
r
S
[
45,46]. In particular, in the photocatalytic oxidation of PFOA a new
chemical route has been identified for the synthesis of carbonyl
difluoride, COF
r
S
S
2
.
because it represents the adsorptive affinity of a substrate on the
catalyst surface. Differently, in the modified L-H model the light
2
. Materials and methods
.1. Materials
Perfluorooctanoic acid (purity 96% – from Sigma–Aldrich ) was
r S
intensity can affect both kinetic constants (k , K ), as reported in the
literature [49,50].
2
The effects on PFOA photocatalytic oxidation due to both
oxygen excess and deficiency in the reaction environment were
evaluated by running and comparing the results of specific kinetic
tests; in particular, PFOA degradation trends were monitored in
1
used as received. PFOA is soluble in water (9.5 g/L) and its critical
micelle concentration (CMC) is 7.80 ꢂ 10ꢀ mol/L at 25 8C [47]. Ti-
tanium dioxide P-25 (75% Anatase, 25% Rutile) was supplied by
Evonik and it was tested as a titanium-based photocatalyst. The
coexistence of anatase and rutile in commercial P-25 causes the
catalyst photoactivity to be enhanced if compared to pure anatase
3
atmospheric conditions (Air test), in the presence of a constant O
flux (O ¼ 7 NL=h) and in the presence of a constant N
test ꢀ F
flux (F
2
2
O
2
2
1
N
¼ 7 NL=h). The latter case comprehended two different
2
oxygen starvation tests: PFOA degradation under N
air-saturated initial solution (N test) and with a N
reaction environment (N2sat test – PFOA solution fluxed with N
12 h until saturation, then F
2
flux with an
-presaturated
for
2
2
[
48]. The presence of small rutile crystallites, in fact, creates a
2
structure characterized by a more stable charge separation,
slowing recombination reactions on anatase; moreover, the
smaller band gap of rutile extends the useful range of photoactivity
N
¼ 7 NL=h during the kinetic test).
2
Dark and photolysis tests were also conducted (Table S.I.9 and
Table S.I.10, respectively).

M. Sansotera et al. / Journal of Fluorine Chemistry 179 (2015) 159–168
161
ꢀ
In order to verify the deactivating effect of fluoride ions toward
[M–H] ions for all expected degradation intermediates. Primary
ions monitored for PFOA, perfluoroheptanoic acid (PFHpA), per-
fluorohexanoic acid (PFHxA), perfluoropentanoic acid (PFPeA),
perfluorobutanoic acid (PFBA), perfluoropropionic acid (PFPrA)
and trifluoroacetic acid (TFA) determinations were 413, 363, 313,
ꢀ
TiO
2
, an additional test (F test) was performed by adding
potassium fluoride ([KF]
0
= 120 mM) to the initial PFOA solution.
In these conditions PFOA photoabatement was not observed, thus
confirming the effects due to the fluoride-promoted deactivation
of the catalyst (see Table S.I.11 in the Supporting Information for
more details).
ꢀ
263, 213, 163 and 113 [M–H] , respectively (see Appendix A).
Samples of the reaction mixture collected at different reaction times
were diluted in deionized water (1:10) and injected in the HPLC–MS
with 2 mM ammonium acetate/methanol as the mobile phase
2.3. Detection of carbonyl difluoride
starting at 10% methanol, at a flow rate of 200
mL/min. The gradient
The intent of this test was the detection of COF
2
, hypothesized
increased to 90% methanol at 5 min; before reverting to original
conditions at 20 min, the gradient decreased to 80% methanol at
15 min. Column temperature was maintained at 40 8C. HPLC–MS
spectrum of Air Test sample collected after 7 h and the extractions
of primary ions due to PFOA and degradation PFCA intermediates
are reported as example in the Supporting Information (Fig. S.I.1).
as a specific by-product in the PFOA degradation pathway based on
the -scission reaction. The photocatalytic apparatus described in
b
Section 3.2 was connected to an on-line Thermo Nicolet 380 FT-IR
3
set with a PTFE gas-tight IR cell (volume 8 cm , length 10 cm)
equipped with CaF
neutralized in a saturated NaHCO
[PFOA] = 4 mM) was dissolved in a high boiling point PFPE-
2
windows. The outlet gases of the cell were
3
aqueous solution. PFOA
1
9
(
0
2.6. Fluorine-19 nuclear magnetic resonance ( F NMR) spectroscopy
1
based solvent (Galden HT-170) and this solution was introduced
in the reactor. The use of aqueous solutions in the photoabatement
1
9
F NMR tests were performed on a Bruker 500 Ultrashield
of PFOA did not allow the detection of COF
2
because of its rapid
spectrometer operating at 470.30 MHz and 305 K, in order to
evaluate the trend of PFOA signals. In particular, for each kinetic
test, the samples of solution collected from the photoabatement
reactor at the beginning and at different specific times were
hydrolysis. Titanium dioxide P-25 (0.66 g/L) was used as a
2
photocatalyst and the UV lamp (75 W/m ) was placed beside
the reactor. Oxygen (F
O
¼ 7 NL=h) was continuously fluxed in the
2
1
9
reactor. FT-IR analysis was used to monitor the composition of
the outlet gases and to verify the formation of COF , by comparing
due to the C 55 O stretching
Table S.I.12). The concentration of COF (expressed in parts per
analyzed by
2
F NMR, using D O as solvent. PFOA integral
2
calculation allowed the detection of degradation products by
considering the ratio between the integrals of the peaks ascribable
ꢀ
1
its typical IR signal at 1928 cm
(
2
to CF
perfluorinated acids; see Appendix A) and the integrals of the
peaks ascribable to CF signals.
3 2 7
groups, present in all the degradation by-products (C –C
million by volume, ppmV) was calculated on the basis of its
reported molar absorptivity at 10,110 cm/mmol [41]. A portion of
2
the outlet gases was bubbled in a NMR tube containing toluene-d
8
19
cooled in a dry ice-acetone bath at ꢀ78 8C [51]. F NMR spectrum
was recorded on a Bruker 500 Ultrashield spectrometer operating
at 470.30 MHz and 305 K.
2.7. X-ray photoelectron spectroscopy (XPS)
X-ray photoelectron spectroscopy spectra were obtained by
using an M-probe apparatus (Surface Science Instruments). The
1
Chemical stability test of pure Galden HT-170 toward UV-
1
activated TiO
were also performed under O
and S.I.14 in the Supporting Information for more details).
2
and dark test of a PFOA solution in Galden HT-170
source was monochromatic Al Ka radiation (1486.6 eV). A spot size
2
flux (F
O
¼ 7 NL=h – see Tables S.I.13
of 200 m and pass energy of 30 eV were used. 1s level
m
m ꢂ 750
m
2
hydrocarbon-contaminant carbon was taken as the internal
reference at 284.6 eV. Fittings were performed by using pure
Gaussian peaks, Shirley’s baseline, and without any constraints. X-
ray photoelectron spectroscopy analysis was performed to study
2.4. TOC analysis and ion chromatography
Total Organic Carbon (TOC) analysis allowed to evaluate and to
the composition of the photocatalyst surface after the photode-
1
monitor the trend of the carbon content of the solution. TOC
2
gradation of a PFOA solution in Galden HT-170. The TiO sample
1
analyses were performed with a Shimadzu TOC 5000 A with a
for XPS analysis was obtained by centrifuging and filtering through
a 0.45 m polycarbonate membrane the surfactant solution at the
combustion/non-dispersive infrared (NDIR) gas analysis method.
The total organic carbon concentration of the PFOA solution at
different photodegradation times was calculated automatically by
comparing the sample with a calibration curve obtained from PFOA
solutions at defined concentrations. In the kinetic study of PFOA
degradation, the data obtained by TOC analysis were considered
m
end of the reaction (15 h); the sample was then dried in inert
atmosphere for 24 h and analyzed. High resolution XPS analyses in
the typical zone of C 1s (as reference) and Ti 2p were performed.
3. Results and discussion
0
for the calculation of C/C ratios. The PFOA degradation generated
fluoride ions in solution and their concentration was monitored by
Ion Chromatography (IC) with a Metrohm 883 Basic IC Plus.
3.1. Photocatalytic oxidation of perfluorooctanoic acid
The solubility of perfluorocarboxylic acids can be approximated
to their critical micelle concentration (CMC). In particular, PFOA is
characterized by a CMC of 0.0078 M and above this value PFOA
molecules tend to aggregate in micelles of average dimension
around 100 nm, originating a colloid [47]. Below the CMC, PFOA
molecules are expected to exhibit lower chemical stability than
2
.5. High-Performance Liquid Chromatography with Mass
Spectrometry (HPLC–MS)
Analytes separation was performed by using an Agilent
100 Series HPLC Value System, consisting of a quaternary pump,
1
vacuum degasser and auto sampler. The instrument was equipped
2
PFOA aggregates toward the coupled effect of TiO and UV light
1
1
with a Lichrocart 55-4 Purospher STAR RP-18 end capped
m) supplied by Merck KGaA. For
[28]. On the basis of the optimizations achieved in previous studies
[27,28], the general experimental conditions were maintained by
column (55 ꢂ 4.0 mm i.d., 3
m
quantitative determination, the chromatographic system was
interfaced to a Bruker Esquire 3000 Plus quadrupole ion trap mass
spectrometer (Bruker Daltonics) operating in negative electrospray
mode. Instrumental parameters were optimized to transmit the
0
setting PFOA initial concentration at 0.0040 M ([PFOA] < CMC),
2
2
TiO content at 0.66 g/L and UV-lamp power at 75 W/m . Different
kinetic tests were realized in order to evaluate the effects of oxygen
excess or defect in the reaction environment. PFOA degradation

1
62
M. Sansotera et al. / Journal of Fluorine Chemistry 179 (2015) 159–168
trends were monitored in static conditions with atmospheric
concentration of O (Air test) and in the presence of a constant O
flux (O ¼ 7 NL=h) in order to obtain an O -enriched
test ꢀ F
environment. Two kinetic tests were also run under N flux in O
starvation conditions, either in the presence of residual oxygen
traces due to the air-saturated initial solution (N test) or in the
complete absence of O obtained by pre-saturating the PFOA
solution with N before starting the photocatalytic experiment
2sat test – see Section 2.2 for details). The surfactant abatement
trends were evaluated by monitoring at different times the
mineralization percentage of the treated solutions by TOC analysis,
the fluoride contents by IC analysis (Tables S.I.1, S.I.3, S.I.5 and S.I.7)
and the degradation intermediates concentrations by HPLC–MS
conditions during PFOA photodegradation was confirmed. More-
over, no PFOA abatement was observed by working in the presence
2
2
2
O
2
2
of TiO as photocatalyst (0.66 g/L) without UV irradiation (dark test,
2
2
2
2
Table S.I.9) or by working just under UV irradiation (75 W/m )
without photocatalyst (photolysis test, Table S.I.10), in accordance
to the literature [24]. The absence of any PFOA abatement in
2
2
the dark test confirmed also that PFOA adsorption on the TiO
surface was negligible.
2
2
(
N
PFOA degradation intermediates were identified and
quantified by HPLC–MS analysis. In the Air test, the concentration
trends of the degradation intermediates in solution followed a
well-defined order: the higher the molecular weight of the
intermediate, the higher its presence in solution (PFHpA > PFH-
PFHxA > PFPeA > PFBA > PFPrA > TFA) (Fig. 2A and B). PFHpA
concentration increased up to a maximum value (0.632 mM)
followed by an evident decrease. This concentration trend can be
(Tables S.I.2, S.I.4, S.I.6 and S.I.8). PFOA concentrations expressed as
ppm of carbon ascribed to PFOA and measured by HPLC–MS were
reported for benchmarking purposes with the experimental TOC
concentrations (Fig. 1-A). PFOA degradation kinetic data were
related to the C
n
! Cnꢀ1 chain length degradation pathway. It is
2
interpolated by pseudo-first order curves (R > 0.96) which
important to notice that the shorter-chain acids formation during
PFOA decomposition observed at the beginning of the reaction and
the presence of TFA and PFPrA after 30 min (Fig. 2B) cannot be
ascribed to the generally accepted photo-redox consecutive
enabled the comparison of the apparent rate constants, kapp
(
Fig. 1B).
In the Air test, TOC content and PFOA concentration resulted to
be constantly decreasing along the whole experiment and a
substantially constant gap can be observed (Fig. 1A). Considering
C
n
! Cnꢀ1 chain length decrease mechanism [29]. These findings
can be justified by an alternative decomposition mechanism based
on the elimination of COF as a decomposition product induced by
-scission reactions of oxyradicals, in competition with the
the O
2
test, the total mineralization was very rapid during the first
2
4
h of abatement and after 7 h PFOA appeared to be almost
b
completely degraded: [PFOA]7h/[PFOA]
0
= 0.034. After 4 h, the
stepwise chain length decrease [28]. Overall, the concentration
trends of PFOA degradation intermediates represented an indirect
ꢀ
mineralization rate clearly decreased, probably due to the F -
promoted chemical modifications of the catalyst surface [27]. The
gap between the trends of TOC content and PFOA degradation was
proof of the coexistence of two degradation pathways. In the O
2
test, PFHpA maximum concentration was comparable with the one
observed in the Air test, as well as the final concentrations of PFHxA
and PFPeA (Fig. 2C and D). On the contrary, PFPrA and TFA were
2
substantially constant after 7 h. The N2sat test was run in N pre-
saturated reaction environment and a very low decrease in the TOC
content was observed (Fig. 1A): at the end of the test mineraliza-
tion was slightly higher than 2%. Differently, the residual oxygen
definitely more concentrated at the end of the O
2
test than in the Air
test. In the presence of O excess, PFOA decrease appeared to be
2
content present in the N
2
test due to the air-saturated initial
definitely more rapid and higher concentrations of short chain
acids were detected. These results suggested that a PFOA oxidation
solution raised the final mineralization to more than 50% and the
PFOA abatement was about 71% (Fig. 1A). These results are in
complete agreement with the expected behavior of the system by
supporting the enhancing role of oxygen in the reaction mecha-
nism of PFOA photodegradation.
The linearization of PFOA degradation data are reported in
Fig. 1B for all the experimental conditions tested, along with the
calculated apparent kinetic constant. The apparent kinetic
mechanism alternative to the C
degradation should be considered, especially in the presence of
appreciable O concentrations. In the N test (Fig. 2E and F), the
n
! Cnꢀ1 stepwise chain length
2
2
formation of shorter-chain acids resulted greatly limited: [PFPrA]-
max and [TFA]max were 0.008 and 0.002 mM, respectively. In fact,
in O
generally accepted stepwise C
mechanism and the PFOA concentration decreased constantly with
gradual increase in C –C acids content and relatively low
2
starvation, the PFOA abatement appeared to follow the
n
! Cnꢀ1 chain length decrease
ꢀ
1
constant obtained in the O
twice the one calculated in the Air test (kAir = 0.1715 h ) (Fig. 1-B).
As expected, tests run in the presence of O starvation (N test
and 2sat test) allowed lower rates of PFOA removal
2
test (k
O
¼ 0:3346 h ) was almost
2
ꢀ1
7
6
2
2
formation of shorter chain acids. In N2sat test, the very limited
PFOA abatement (Fig. 1) and the related small increase in the PFOA
degradation intermediates concentrations (Fig. 2 and H) were
probably due to traces of oxidant species in the solution ascribable
N
ꢀ
1
ꢀ1
(
k
N
¼ 0:0560 h ; k 2
N
sat ¼ 0:0078 h ). From the comparison
2
of the calculated kapp, the importance of working in O -enriched
2
Fig. 1. Concentration trends of TOC data (solid lines) compared to the ppm of carbon corresponding to PFOA concentration monitored by HPLC–MS (dashed lines) for the
different kinetic tests (O , Air, N and N2sat test) (A); linearization of PFOA degradation data obtained in the different kinetic tests (O , Air, N and N2sat test) (B).
2
2
2
2

M. Sansotera et al. / Journal of Fluorine Chemistry 179 (2015) 159–168
163
0
0
0
.035
.028
.021
A
0.7
PFHpA
PFHxA
PFPeA
PFBA
B
PFPrA
TFA
0
0
0
0
0
.5
.4
.3
.1
.0
0.014
0.007
0.000
0
200
400
600
800
0
0
0
200
400
Time (min)
600
600
600
800
800
800
Time (min)
D 0.035
C
0.7
PFHpA
PFPeA
PFHxA
PFBA
PFPrA
TFA
0
0
0
0
.028
.021
.014
.007
0
0
0
0
0
.5
.4
.3
.1
.0
0.000
0
200
400
Time (min)
600
800
200
400
Time (min)
0
0
0
0
.035
.028
.021
.014
E 0.7
PFHpA
PFHxA
PFPeA
PFBA
F
PFPrA
TFA
0
0
0
0
0
.5
.4
.3
.1
.0
0.007
0.000
0
200
400
Time (min)
600
800
200
400
Time (min)
H 0.035
G 0.7
PFHpA
PFHxA
PFPeA
PFBA
PFPrA
TFA
0
.028
0
0
0
0
0
.5
.4
.3
.1
.0
0.021
0
0
0
.014
.007
.000
0
200
400
600
800
0
200
400
600
800
Time (min)
Time (min)
Fig. 2. Concentration trends of PFOA degradation intermediates (PFHpA, PFHxA, PFPeA, PFBA, PFPrA, TFA) in Air test (A, B), in O
2
test (C, D), N
2
test (E, F) and N2sat test (G, H).
5
2 6
of the TiO catalyst and the undesired production of SiF . Fluoride
to residual O
water adsorbed onto TiO
suggest that the use of an external source of oxygen can promote
the PFOA degradation.
PFOA degradation is expected to produce fluoride ions due to
several routes such as dehydrofluorination of primary perfluori-
2
or to the reaction between photogenerated holes and
particles [52]. These evidences overall
2
concentration in solution increased linearly during the Air test
ꢀ
until around 9 h of photoabatement; then F increase was less
pronounced (Fig. S.I.2B). At the end of the treatment, the detected
ꢀ
fluoride concentration was 29 mM, corresponding to a F yield of
48%. Fluoride release in O
case of Air test (Fig. S.I.3B): the F concentration reached 25 mM
after 4 h which was about 2.5 times the F content observed in the
2
test was definitely quicker than in the
ꢀ
nated alcohols and hydrolysis of both COF
2
and perfluoroacyl
ꢀ
fluoride intermediates. Overall, fluoride release can be considered
as a complementary marker of PFOA mineralization; however,
differences in the balance between fluoride yields and mineraliza-
tion degrees can be noticed and should be ascribed to several
contributory causes, such as the volatility of some PFOA
Air test after the same degradation time (10 mM after 4 h); after 7 h
ꢀ
treatment, the F concentration reached a plateau and a maximum
value of 31 mM, slightly higher than the one in the Air test. This
trend can be ascribed to an almost complete deactivation of the
decomposition products, like HF and COF
2
, the partial fluorination
catalyst due to the chemical interactions of fluoride ions with TiO
2

1
64
M. Sansotera et al. / Journal of Fluorine Chemistry 179 (2015) 159–168
surface moieties [27,28]. In N
2
test as well as in N2sat test, a limited
degrees, WAir were always significantly lower than
mineralization of 62%, WAir at 9 h was equal to 2.9, while
was 3.3; more markedly, at mineralization of 80% WAir at 15 h was
equal to 1.6, that is definitely lower than 2.3 ( at 7 h). A similar
trend was observed in Air and N tests: at mineralization of 52%, WAir
at 7 h was found equal to 3.7 while at 15 h was 3.3.
On the basis of these results it is possible to hypothesize that in
the presence of a constant O feed (O test) the degradation of PFOA
preferentially follows the -scission pathway. On the contrary, in
starvation conditions (N test), the photo-redox C
# (Table 1): at
O
2
fluoride release was observed (Figs. S.I.4B and S.I.5B): at the end of
#
O
at 4 h
2
ꢀ
ꢀ
the tests (after 15 h), [F ] were about 6 and 3 mM, respectively; F
yields were around 10% and 5%, respectively. These results confirm
that the photocatalytic degradation of fluorinated surfactants in
solution is significantly influenced by the presence of dissolved
oxygen.
#
O
2
2
#
N
2
2
2
1
9
3
.2. F NMR results
b
O
2
2
n
! Cnꢀ1 chain
Referring to the labeled formula CF3(h
)
CF2( )CF2( )CF2(
)
d
length mechanism appears to be preponderant. In the case of PFOA
degradation in the presence of naturally dissolved oxygen (Air test),
z
e
1
9
CF2(
)
CF2(
)
a
)
CF2( COOH, the assignments of PFOA in the F NMR
g
b
spectra (Fig. S.I.6) were in agreement with literature:
d
= ꢀ81.9
a balanced coexistence of
is possible.
b-scission and photo-redox pathways
(
(
3 F, F ), ꢀ127.1 (2 F, F ), ꢀ123.8 (2 F, F ), ꢀ123.1 (2 F, F ), ꢀ122.9
h
z
e
d
2 F, F ), ꢀ124.2 (2 F, F ), ꢀ118.5 (2 F, F ) ppm [53,54]. The peaks of
It is also interesting to compare the CF3(h
different PFOA mineralization (Fig. 3A). In the spectra of Air test at
15 h and O test at both 7 and 15 h, the well-resolved triplet of
triplets of the PFOA CF3(h
signal at ꢀ81.86 ppm varied in multiple
)
signals obtained at
g
b
a
shorter perfluorinated acids (C7ꢀ3) have chemical shifts located in
the immediate vicinity of PFOA signals, except for trifluoroacetic
acid (TFA) that is characterized by a single signal at ꢀ76.55 ppm
2
)
1
9
(generally used as a calibration standard) [28,55,56]. F NMR
signals resulting from the partial overlapping of several contribu-
tions: a triplet at ꢀ82.15 ppm due to PFPrA, a triplet of triplets at
ꢀ82.03 ppm due to PFPeA and a multiplet in the range between
ꢀ81.86 and ꢀ81.92 ppm due to overlapping triplets of triplets
ascribable to PFOA, PFHpA, PFHxA and PFBA. This transition
confirmed the coexistence of comparable amounts of different
perfluorocarboxylic acids (C8ꢀ3). In addition, the presence of TFA
spectra of PFOA solutions collected from the photoabatement
reactor at different times during the kinetic tests (Air, O , N2sat and
tests) were recorded in order to confirm the variation in the
chemical composition of the samples. The integrals of the signals
ascribable to CF2( - ) and CF3( were calculated for each spectra
2
N
2
a z
h
)
1
9
and the corresponding ratios (W) between the sum of CF2(a-z)
F
NMR integrals and the sum of CF3(
)
integrals were also evaluated
(C
S.I.7–11). Conversely, in the spectra of N
7 and 15 h, the CF3(h signal of PFOA remained almost unchanged
(Fig. S.I.12 and S.I.13); in N test at 15 h a slight shoulder at
2
) was also proved by its distinctive signal at ꢀ76.55 ppm (Fig.
h
P
P
(
Table 1): W = (CF2( - ) integrals)/ (CF3(
h
)
integrals). The direct
2
and N2sat tests at both
a z
mineralization of PFOA (Eq. (2)) progressively reduces the
concentrate of the starting acid, formally avoiding the formation
of shorter chain species and maintaining constant the ratio W:
)
2
ꢀ81.92 ppm can be detected. Comparatively, in these tests it was
a z
possible to notice a complementary trend in the CF2( - ) signals
CF
3
ðCF
2
Þ
6
COOH þ 7=2 O
2
þ 7 H
2
O ! 8 CO
2
þ 15 HF
(2)
(
from ꢀ118 to ꢀ128 ppm) obtained at different PFOA mineraliza-
tion (Fig. 3B): in the spectra of Air test at 15 h and O test at both
and 15 h, CF signals almost disappeared and new weak CF
signals appeared at ꢀ118.7 (triplet; CF2( of PFBA), ꢀ119.5
2
On the contrary, if the PFOA decomposition proceeds across
7
2
2
19
intermediates, shorter chain acids are generated and the F NMR
integrals ascribable to CF moieties in the chains decrease
a
)
2
(
quartet; CF2(
)
of PFPrA), ꢀ124.5 (multiplet; CF2(
b
)
of PFPeA) and
a
independently from the terminal CF
is expected to decrease.
3
integrals; thus, the ratio W
ꢀ
125.2 ppm (multiplet; CF2(
)
of PFBA), ascribable to low
b
concentration of shorter perfluorocarboxylic acids [56–62]. In N
2
As reported in Table 1, the calculated W value in the starting
solution was equal to 4.0. The comparison of W values at similar
mineralization degrees in different tests reveals the relative
incidence of the two alternative degradation pathways: in particu-
lar, low W values can be ascribed to a higher influence of the
test at 15 h and in Air test at 7 h, only a broadening of the intense
signal of the CF2( - ) peaks was observed.
a z
3.3.
b-Scission of perfluorooxyradicals and carbonyl difluoride
detection
photo-redox C
ments characterized by high W values preferably follow the
scission pathway. In Air and O tests, the comparison of W values
n
! Cnꢀ1 stepwise mechanism; conversely, experi-
b-
The PFOA degradation passes through the formation of oxygen-
centered perfluororadicals, as already reported in the literature
28]. The commonly accepted mechanisms foresee that these
2
highlighted a well-defined trend; in fact at similar mineralization
[
radicalic intermediates evolve by producing unstable perfluori-
nated alcohols, which further decompose to perfluoroacyl fluorides
by eliminating hydrofluoric acid [63]. However, the
reaction of perfluorinated oxyradicals (C O) is thermodynamically
favored with a low kinetic barrier; shorter carbon-centered
perfluoroalkyl radicals (Cnꢀ1) and COF are formed as reaction
products [18]. Perfluoroalkyl radicals continue the degradation
pathway and COF is rapidly hydrolyzed in aqueous media by
generating CO and HF. To demonstrate this pathway, the
hydrolysis of COF should be inhibited by setting the experiment
in the absence of water and allowing the release of COF from the
b-scission
Table 1
n
Comparison of W values at different decomposition time for Air, O
kinetic tests.
2
, N
2
and N2sat
2
a
Test
Time (h)
W (ꢀ)
Mineralization (%)
Air test
0
7
9
4.0
3.7
2.9
1.6
4.0
3.3
2.3
4.0
3.9
3.3
4.0
4.0
4.0
0
50
62
80
0
2
2
2
1
5
0
4
7
0
7
5
0
7
5
O
2
test
test
2
69
78
0
reaction environment in gaseous phase. Therefore, a specific test
was defined in order to achieve the PFOA degradation in absence of
N
N
2
2
water and to avoid the COF hydrolysis. A high boiling point
30
54
0
1
perfluoropolyether (Galden HT-170, b.p. = 170 8C) was chosen as
a solvent because of its solvating properties on PFOA (ꢅ1.9 g/L),
inertness to radicals, stability toward UV-light and low volatility.
The remaining experimental parameters of the test were main-
1
1
2sat test
1
2
a
%
mineralization calculated on the basis of TOC data.
2
tained consistent with the O test by setting PFOA initial

M. Sansotera et al. / Journal of Fluorine Chemistry 179 (2015) 159–168
165
A
B
O
2
test - 15
h
O
2
test
-
-
15
h
h
O
2
test - 7
h
O
2
test
-
7
Air test
Air test
-
-
15
h
Air test
15
h
h
Air test
-
7
7
h
N
2
test
-
15
h
h
N
2
test - 15
h
N
2
test
-
-
7
2
N test - 7 h
N
2sat test
-
15
h
h
N
2Sat test - 15
h
N
2sat test
7
N2Sat test - 7 h
-
81.9
- 82.0
- 82.1
[ppm]
-120
-122
-124
(A) and CF2(
-126 [ppm]
1
9
Fig. 3. Comparison of F NMR spectra recorded at 7 and 15 h for Air, O
2
, N2sat and N
2
kinetic tests: zooms in the CF3(
h
)
a
-
z
)
(B) regions.
concentration at 4 mM, TiO
2
content at 0.66 g/L, UV-lamp power at
In the PFPE-solvent PFOA was completely degraded after
2
7
5 W/m and oxygen flux at 7 NL/h. Oxygen was continuously
approximately 12 h and the maximum COF
flux resulted equal to 232 ppmV. The COF
2
concentration in the
2
concentration trend
fluxed in the reactor in order to enhance the concentration of
perfluorinated peroxyradicals and consequently to facilitate the
over time revealed the rapid and almost complete decomposition
formation of COF
the perfluoropolyether solvent started, the outlet gases from the
reactor were analyzed by FT-IR spectroscopy and the production of
2
. As soon as the PFOA photocatalytic oxidation in
of the PFOA through the
the production of COF
b
-scission pathway (Fig. 4). In addition,
and its removal from the reaction
2
environment, due to stripping action by the oxygen flow,
reduced the release of fluoride ions in the solution. Thus, TiO
2
deactivation due to fluoride ions was inhibited by performing
PFOA photooxidation in a PFPE-based solvent and, even after 15 h
of photoabatement only a minor part of the catalyst turned out to
COF
of COF
stretching was easily recognized in the spectra (Fig. S.I.14) [41]. The
2
was monitored (Fig. 4 and Table S.I.12). The typical IR pattern
ꢀ1
2
with the intense signal at 1928 cm due to the C 55 O
ꢀ1
IR signals of CO
lines in the range 3700–4200 cm were also present. A portion of
2
antisymmetrical stretching at 2350 cm and HF
ꢀ
1
2
be fluorinated. At the end of the test, XPS analysis of TiO catalyst
19
the outlet gases was also analyzed by F NMR spectroscopy and the
(Fig. S.I.16) showed in the Ti 2p region two couples of peaks which
can be attributed to pure titanium dioxide (at 458–464 eV) and to
titanium fluorides with low fluorination degree (at 460–466 eV).
On the basis of the deconvolutions, the amount of
these fluorinated species was estimated as 30% of the surface
composition.
spectrum revealed an evident signal at ꢀ21.5 ppm ascribable to
2
COF (Fig. S.I.15) [51]. While the concentration trends of PFOA
degradation intermediates indirectly suggest the existence of two
competitive degradation pathways, the detection of COF can be
considered as a direct experimental evidence of the
route in the PFOA degradation mechanism.
2
b-scissions
3.4. Suggested mechanism for PFOA photocatalytic oxidation
2
2
1
1
50
00
50
00
The experimental results discussed in this work allowed new
insights on the PFOA degradation mechanism. In particular, the
role of oxygen, the concentration trends of the degradation
intermediates and the detection of COF
2
offer a new interpretation
of the photocatalytic oxidation of PFOA.
It is commonly recognized that the PFOA degradation mecha-
nism is initiated with the excitation of titanium dioxide caused by
the irradiation of UV light [29,36,64]; after being excited by UV
radiation (Fig. 5 – reaction (a)), TiO
dissociated PFOA in water (CF (CF
2
accepts one electron from
ꢀ
3
2
)
6
COO ), generating PFOA
5
0
radical (Fig. 5 – reaction (b)) [29,36,64]. As clarified by various
papers, PFOA decomposition starts in correspondence to the
carboxylic function that undergoes a Kolbe decarboxylation (Fig. 5
– reaction (c)) [29,65–67].
0
0
200
400
600
800
Time (min)
h
y
ꢆ
þ
TiO
2
! TiO ðh þ eꢀÞ
(a)
(b)
2
Fig. 4. COF
2
concentration (expressed as parts per million by volume, ppmV) in
outlet gases during the decomposition of PFOA dissolved in a high boiling point
1
h
þ
ꢀ
Þ6COOꢁ
CF
3
ðCF
2
Þ COO ! CF
3
ðCF
2
PFPE-based solvent (Galden HT-170).
6

1
66
M. Sansotera et al. / Journal of Fluorine Chemistry 179 (2015) 159–168
2
Fig. 5. Reaction mechanism of PFOA degradation in the presence of TiO photocatalyst: photo-redox pathway (A – blue arrows); b-scission pathway (B – orange arrows). (For
interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
ꢁ
ꢁ
to ꢀ160 kJ/mol) and originate the corresponding acyl fluorides and
hydrogen fluoride (Fig. 5 – reaction (h)) [66,70]; in the presence
of water, the acyl fluorides hydrolyze to carboxylic acids (Fig. 5 –
reaction (i)) [70–73].
CF
3
ðCF
2
Þ
6
COO ! CF
3
ðCF
2
Þ
5
CF
2
þ CO
2
(c)
The energetically favored reaction of C
formation of primary perfluorinated alcohols by coupling with
hydroxyl radicals produced by water oxidation due to photo-
7
radicals includes the
IV
ꢁ+
CF
3
ðCF
2
Þ
5
CF
2
OH ! CF
3
ðCF
2
Þ
5
CðOÞF þ HF
(h)
(i)
generated holes (Ti OH ) (Fig. 5 – reaction (d)) [68]:
ꢁ
ꢁ
CF ðCF Þ CðOÞF þ H O ! CF ðCF Þ COOH þ HF
CF
3
ðCF
2
Þ
5
CF
2
þ OH ! CF
3
ðCF
2
Þ
5
CF
2
OH
(d)
3
2
5
2
3
2
5
The reactions from (a) to (i) are at the basis of the stepwise
However, as demonstrated by the kinetic tests,
a
high
C
n
! Cnꢀ1 chain length shortening during PFOA degradation. The
concentration of oxygen in the reaction environment promotes
the coupling between C radicals and O by inducing the formation
of peroxyradicals (Fig. 5 – reaction (e)) [69]:
evidences obtained from the concentration trends of the degrada-
7
2
2
tion intermediates and the detection of COF prove the existence of
another reaction pathway, where the oxyradical formed in reaction
ꢁ
ꢁ
CF
3
ðCF
2
Þ
5
CF
2
þ O
2
! CF
3
ðCF
2
Þ
5
CF
2
OO
(e)
(f) evolves by eliminating COF through monomolecular
b
-scission
2
and consequent generation of a Cnꢀ1 radical [4,74,75]:
The coupling of two peroxyradicals allows the production of
ꢁ
ꢁ
oxyradicals (Fig. 5 – reaction (f)) that, in the presence of the surface
excited electrons of TiO and water, generate an unstable primary
perfluorinated alcohol (Fig. 5 – reaction (g)) [69,70].
CF ðCF Þ CF O ! CF ðCF Þ CF þ COF2
(j)
3
2
5
2
3
2
4
2
2
The carbon-centered perfluorinated radical is redirected at
the step of reaction (e), while fluorophosgene is quenched in
the aqueous environment generating hydrogen fluoride and
carbon dioxide [70]. It is remarkable that the mechanism based
ꢁ
ꢁ
2
CF
3
ðCF
2
Þ
5
CF
2
OO ! 2 CF
3
ðCF
2
Þ
5
CF
2
O
þ O
2
(f)
ꢀ
e
OH þ OHꢀ
on b-scissions can promote an almost complete PFOA
ꢁ
CF
3
ðCF
2
Þ5CF
2
O þ H
2
O ! CF
3
ðCF
2
Þ5CF
2
(g)
decomposition without the formation of perfluorinated acids as
intermediates. Thus, the key steps of the complete oxidation
mechanism are represented by the two competing reactions: the
Alternatively, peroxyradicals can react with molecular oxygen
originating oxyradicals and releasing ozone that can react with C
radicals, generating further oxyradicals [18].
Primary perfluorinated alcohols produced in reactions (d) and
g) are widely reported as thermodynamically unstable (from ꢀ80
7
bimolecular reduction of the oxyradical activated by TiO
(Fig. 5 – reaction (g)) and the monomolecular -scission
generating COF and Cnꢀ1 radical (Fig. 5 – reaction (j)).
2
surface
b
(
2

M. Sansotera et al. / Journal of Fluorine Chemistry 179 (2015) 159–168
167
4
. Conclusions
References
The photocatalytic oxidation of PFOA promoted by titanium
[1] J.G. Drobny, Technology of Fluoropolymer, second ed., CRC Press, Boca Raton,
009.
2
dioxide was studied by testing the reaction in different conditions:
static air, oxygen flux, nitrogen flux and pre-saturated nitrogen
flux. The effects of oxygen excess or defect in the reaction
environment were evaluated in different kinetic tests by measur-
ing mineralization degrees and fluoride releases by TOC analysis
and IC, respectively. Concentration trends of PFOA degradation
[2] W. Navarrini, M.V. Diamanti, M. Sansotera, F. Persico, M. Wu, L. Magagnin, S.
Radice, Prog. Org. Coat. 74 (2012) 794–800.
[3] F. Persico, M. Sansotera, C.L. Bianchi, C. Cavallotti, W. Navarrini, Appl. Catal. B
70 (2015) 83–89.
4] M. Avataneo, W. Navarrini, U. De Patto, G. Marchionni, J. Fluorine Chem. 130
2009) 933–937.
1
[
(
[5] C. Cornelis, W. D’Hollander, L. Roosens, A. Covaci, R. Smolders, R. Van den
Heuvel, E. Govarts, K. Van Campenhout, H. Reynders, L. Bervoets, Chemosphere
1
9
intermediates were also determined by HPLC–MS and F NMR
analyses. In particular, the disclosure of very short chain acids at
the beginning of the reaction suggested the presence a PFOA
oxidation mechanism alternative to the stepwise photo-redox
86 (2012) 308–314.
[
[
6] J.H. Johansson, U. Berger, R. Vestergren, I.T. Cousins, A. Bignert, A. Glynn, P.O.
Darnerud, Environ. Pollut. 188 (2014) 102–108.
7] R.C. Buck, J. Franklin, U. Berger, J.M. Conder, I.T. Cousins, P. de Voogt, A.A.
Jensen, K. Kannan, S.A. Mabury, S.P.J. van Leeuwen, Integr. Environ. Assess.
Manag. 7 (2011) 513–541.
C
n
! Cnꢀ1 pathway: the
elimination. The mechanism based on
appeared to be dominant in O -enriched solutions, while in O
starving conditions the photo-redox C
! Cnꢀ1 chain length
decrease mechanism prevailed. These hypotheses were confirmed
by detecting carbonyl difluoride, COF , specifically produced in the
-scission pathway. An appropriate kinetic test was set for COF
b
-scissions routes mediated by COF
2
b
-scission reactions
[8] E. Smulders, W. Von Rybinski, A. Nordskog, Laundry Detergents. Ullmann’s
Encyclopedia of Industrial Chemistry, seventh ed., Wiley-VCH, Weinheim, 2011.
2
2
-
[9] C.A. Moody, J.A. Field, Environ. Sci. Technol. 34 (2000) 3864–3870.
n
[
10] A. Singh, J.D. Van Hamme, O.P. Ward, Biotechnol. Adv. 25 (2007) 99–121.
[11] G. Kostov, F. Boschet, B. Ameduri, J. Fluorine Chem. 130 (2009) 1192–1199.
[12] G. Boutevin, D. Tiffes, C. Loubat, B. Boutevin, B. Ameduri, J. Fluorine Chem. 134
2
(2012) 77–84.
b
2
[
[
13] G. Marchionni, V. Tortelli, I. Wlassics, V. Kapeliouchko, US 8703889 B2 (2014).
14] M.A. Guerra, K. Hintzer, M. Jurgens, H. Kaspar, A.R. Maurer, G.I. Moore, Z.M.
Qiu, J.F. Schulz, W. Schwertfeger, T. Zipplies, US 0276103 A1 (2007).
15] T. Ishikawa, N. Tsuda, Y. Yamamoto, US 7777075 B2 (2010).
16] J. Hoshikawa, S. Higuchi, Y. Matsuoka, N. Yamagishi, US 7709566 B2 (2010).
17] P.D. Brothers, S.V. Gangal, US 7932333 B2 (2011).
detection by avoiding its easy hydrolysis and a high boiling point
perfluoropolyether as reaction solvent was employed because of
[
[
[
the solvating properties on PFOA. Thus, COF
characterized by F NMR and IR analyses. The production of COF
2
production was
19
2
in a non-aqueous medium also reduced the release of fluoride ions
in the reaction environment and hindered the F -induced catalyst
[18] J. Niu, H. Lin, C. Gong, X. Sun, Environ. Sci. Technol. 47 (2013) 14341–14349.
[19] S.M. Mitchell, M. Ahmad, A.L. Teel, R.J. Watts, Environ. Sci. Technol. Lett. 1
ꢀ
(2014) 117–121.
deactivation. In addition to the confirmation of the mechanism, it
is worth to notice that carbonyl difluoride is an important
industrial intermediate and its synthesis with good selectivity
and high purity was possible starting from PFOA as pioneering
recycling of this banned fluorinated surfactant.
[
[
[
20] J.C. Lin, S.L. Lo, C.Y. Hub, Y.C. Lee, J. Kuo, Ultrason. Sonochem. 22 (2015)
542–547.
21] N. Takeuchi, Y. Kitagawa, A. Kosugi, K. Tachibana, H. Obo, K. Yasuoka, J. Phys.
D: Appl. Phys. 47 (2014) 045203.
22] Y. Lee, S. Lo, J. Kuo, C. Hsieh, Front. Environ. Sci. Eng. 6 (2012) 17–25.
[23] W. Yuan, P. Zhang, J. Environ. Sci. 26 (2014) 2207–2214.
[24] R.R. Giri, H. Ozaki, T. Okada, S. Taniguchi, R. Takanami, Chem. Eng. J. 180
(
2012) 197–203.
Acknowledgments
[
[
25] M.J. Chen, S.L. Lo, Y.C. Lee, C.C. Huang, J. Hazard. Mater. 288 (2015) 168–175.
26] B. Zhao, X. Li, L. Yang, F. Wang, J. Li, W. Xia, W. Li, L. Zhou, C. Zhao,
Photochem. Photobiol. 91 (2015) 42–47.
The authors wish to acknowledge with thanks the generous
support and the valuable interactions induced to this research
in the field of fluorinated materials by the institution of the
Politecnico di Milano/Solvay Fluorine Chemistry Chair. This
work has been supported by MIUR (PRIN 2010–2011, prot.
[
27] M. Sansotera, F. Persico, C. Pirola, W. Navarrini, A. Di Michele, C.L. Bianchi,
Appl. Catal. B 148 (2014) 29–35.
[28] S. Gatto, M. Sansotera, F. Persico, M. Gola, C. Pirola, W. Panzeri, W. Navarrini,
C.L. Bianchi, Catal. Today 241 (2015) 8–14.
29] S.C. Panchangam, A.Y.C. Lin, J.H. Tsai, C.F. Lin, Chemosphere 75 (2009) 654–660.
30] C.D. Vecitis, H. Park, J. Cheng, B.T. Mader, M.R. Hoffmann, Front. Environ. Sci.
Eng. China 3 (2009) 129–151.
[
[
2
010PFLRJR).
[
[
31] J. Liu, S.M. Avenda n˜ o, Environ. Int. 61 (2013) 98–114.
32] Z. Wang, I.T. Cousins, M. Scheringer, K. Hungerb u¨ hler, Environ. Int. 60 (2013)
242–248.
Appendix A. Glossary of acronyms, names (classical and IUPAC)
and condensed chemical formulas for PFCAs
[
[
[
[
[
33] D. Friedmann, C. Mendive, D. Bahnemann, Appl. Catal. B 99 (2010) 398–406.
34] O. Carp, C.L. Huisman, A. Reller, Prog. Solid State Chem. 32 (2004) 33–177.
35] N. Serpone, J. Photochem. Photobiol. A 104 (1997) 1–12.
36] A. Fujishima, T.N. Rao, D.A. Tryk, J. Photochem. Photobiol. C 1 (2000) 1–21.
37] K. Sato, T. Hirakawa, A. Komano, S. Kishi, C.K. Nishimoto, N. Mera, M.
Kugishima, T. Sanoa, H. Ichinose, N. Negishi, Y. Seto, K. Takeuchi, Appl. Catal. B
PFOA Perflurooctanoic acid (or perfluorocaprylic acid)
,2,3,3,4,4,5,5,6,6,7,7,8,8,8-Pentadecafluorooctanoic acid
PHpA Perfluroheptanoic acid (or perfluoroenhantic acid)
,2,3,3,4,4,5,5,6,6,7,7,7-Tridecafluoroheptanoic acid
PFHxA Perflurohexanoic acid (or perfluorocaproic acid)
,2,3,3,4,4,5,5,6,6,6-Undecafluorohexanoic acid
PFPeA Perfluropentanoic acid (or perfluorovalerianic acid)
,2,3,3,4,4,5,5,5-Nonafluoropentanoic acid
PFBA Perflurobutanoic acid (or perfluorobutyric acid)
,2,3,3,4,4,4-Heptafluorobutanoic acid
PFPrA Perfluropropanoic acid (or perfluoropropionic acid)
,2,3,3,3-Pentafluoropropanoic acid
Trifluoroethanoic acid (or trifluoroacetic acid)
,2,2-Trifluoroethanoic acid
CF
3
CF
3
CF
3
CF
3
CF
3
CF
3
CF
3
(CF
(CF
(CF
(CF
(CF
2
2
2
2
2
)
)
)
)
)
6
5
4
3
2
COOH
COOH
COOH
COOH
COOH
2
106 (2011) 316–322.
[
38] S.R. Seagle, in: J.I. Kroschwitz (Ed.), fourth ed., The Kirk-Othmer Encyclopedia
2
of Chemical Technology, vol. 24, John Wiley and Sons, New York, 1997, pp.
1
86–224.
39] N.P. Mellott, C. Durucan, C.G. Pantano, M. Guglielmi, Thin Solid Films 502
2006) 112–120.
2
[
[
(
2
40] P. Pichat, NATO ASI Series, in: E. Pelizzetti, N. Serpone (Eds.), Homogeneous
and Heterogeneous Photocatalysis, vol. 174, Springer, New York/Heidelberg,
1986, pp. 533–554.
2
CF
2
COOH
[41] J. Pacansky, R.J. Waltman, J. Phys. Chem. 95 (1991) 1512–1518.
[42] V. Francesco, M. Sansotera, W. Navarrini, J. Fluorine Chem. 155 (2013)
2–20.
2
TFA
COOH
[43] W. Navarrini, V. Tortelli, A. Russo, S. Corti, J. Fluorine Chem. 95 (1999)
2
27–39.
[
[
44] W. Navarrini, A. Russo, V. Tortelli, J. Org. Chem. 60 (1995) 6441–6443.
45] M. Sansotera, W. Navarrini, M. Gola, C.L. Bianchi, P. Wormald, A. Famulari, M.
Avataneo, J. Fluorine Chem. 132 (2011) 1254–1261.
[
46] W. Navarrini, C.L. Bianchi, L. Magagnin, L. Nobili, G. Carignano, P. Metrangolo,
G. Resnati, M. Sansotera, Diam. Relat. Mater. 19 (2010) 336–341.
47] U. J a¨ rnberg, K. Holmstr o¨ m, B. van Bavel, A. K a¨ rrman, Perfluoroalkylated acids
and related compounds (PFAS) in the Swedish environment – Chemistry,
Sources & Exposure. Report to Swedish Environment Protection Agency, 2006.
48] C. Gambarotti, C. Punta, F. Recupero, T. Caronna, L. Palmisano, Curr. Org.
Chem. 14 (2010) 1153–1169.
Appendix B. Supplementary data
[
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.jfluchem.2015.
[
0
6.019.

1
68
M. Sansotera et al. / Journal of Fluorine Chemistry 179 (2015) 159–168
[
[
[
49] S. Brosillon, L. Lhomme, C. Vallet, A. Bouzaza, D. Wolbert, Appl. Catal. B 78
2008) 232–241.
50] J.B. Heredia, J. Torregrosa, J.R. Dominguez, J.A. Peres, J. Hazard. Mater. 83
2001) 255–264.
[63] R.D. Chambers, in: R.D. Chambers (Ed.), Fluorine in Organic Chemistry,
Blackwell Publishing Ltd, Oxford, 2004, pp. 236–295.
[64] A. Mills, S. Le Hunte, J. Photochem. Photobiol. A 108 (1997) 1–35.
[65] H. Hori, E. Hayakawa, H. Einaga, S. Kutsuna, K. Koike, T. Ibusuki, H. Kiatagawa,
R. Arakawa, Environ. Sci. Technol. 38 (2004) 6118–6124.
(
(
51] P. Sˇ vec, A. Eisner, L. Kol a´ rˇ ov a´ , T. Weidlich, V. Pejchal, A. R u˚ zˇ i cˇ ka, Tetrahedron
Lett. 49 (2008) 6320–6323.
52] B. Ameduri, Chem. Rev. 109 (2009) 6632–6686.
53] A.H. Karoyo, A.S. Borisov, L.D. Wilson, P. Hazendonk, J. Phys. Chem. B 115
[66] Y. Wang, P. Zhang, J. Hazard. Mater. 192 (2011) 1869–1875.
[67] H. Hori, A. Yamamoto, E. Hayakawa, S. Taniyasu, N. Yamashita, S. Kutsuna,
Environ. Sci. Technol. 39 (2005) 2383–2388.
[
[
(
2011) 9511–9527.
[68] C. Gong, X. Sun, C. Zhang, X. Zhang, J. Niu, Int. J. Mol. Sci. 15 (2014) 14153–14165.
[69] H. Lin, J. Niu, S. Ding, L. Zhang, Water Res. 46 (2012) 2281–2289.
[70] S. Kutsuna, H. Hori, Int. J. Chem. Kinet. 39 (2007) 276–288.
[71] C. Kormann, D.W. Bahnemann, M.R. Hoffmann, Environ. Sci. Technol. 25
(1991) 494–500.
[72] M. Sansotera, W. Navarrini, G. Resnati, P. Metrangolo, A. Famulari, C.L. Bianchi,
P.A. Guarda, Carbon 48 (2010) 4382–4390.
[73] W.J. De Bruyn, J.A. Shorter, P. Davidovits, D.R. Worsnop, M.S. Zahniser, C.E.
Kolb, Environ. Sci. Technol. 29 (1995) 1179–1185.
[
[
[
[
[
54] A.A. Ribeiro, J. Fluorine Chem. 83 (1997) 61–66.
55] W.R. Dolbier, Guide to Fluorine NMR for Organic Chemists, Wiley, Hoboken, 2009.
56] G. Filipovich, G.V.D. Tiers, J. Phys. Chem. 63 (1959) 761–763.
57] D.O. Graham, W.B. McCormack, J. Org. Chem. 31 (1966) 958–959.
58] C. Dapremont-Avignon, P. Calas, A. Commeyras, C. Amatore, J. Fluorine Chem.
5
1 (1991) 357–379.
59] D.A. Ellis, K.A. Denkenberger, T.E. Burrow, S.A. Mabury, J. Phys. Chem. A 108
2004) 10099–10106.
[
(
[
[
60] A.A. Ribeiro, K. Umayahara, Magn. Reson. Chem. 41 (2003) 107–114.
61] N. Ilayaraja, A. Manivel, D. Velayutham, M. Noel, J. Appl. Electrochem. 38
[74] A.M.B. Giessing, A. Feilberg, T.E. M o¨ gelberg, J. Sehested, M. Bilde, T.J.
Wallington, O.J. Nielsen, J. Phys. Chem. 100 (1996) 6572–6579.
[75] M. Sansotera, W. Navarrini, L. Magagnin, C.L. Bianchi, A. Sanguineti, P.
Metrangolo, G. Resnati, J. Mater. Chem. 20 (2010) 8607–8616.
(2008) 175–186.
[
62] Z. Liu, J.D. Goddard, J. Phys. Chem. A 113 (2009) 13921–13931.