A photoionization study of trifluoromethanol, CF3OH, trifluoromethyl hypofluorite, CF3OF, and trifluoromethyl hypochlorite, CF3OCl

Article abstract of DOI:10.1063/1.474017

CF₃OH, an important and controversial by-product of atmospheric decomposition of CF₃CFH₂ (HFC-134a) and other hydrofluorocarbons, has been examined by photoionization mass spectrometry. The ionization onset is characterized by a broad Franck-Condon distribution, arising primarily from a substantial elongation of the C-O bond upon ionization. An upper limit to the adiabatic ionization potential (IP) of ≤ 13.08 ± 0.05 eV has been established. The appearance potentials (APs) of the first two fragments have been accurately determined by fitting with appropriate model functions as AP₀(CF₂OH⁺/CF ₃OH)≤13.830±0.005 eV and AP₀(CF⁺₃/CF ₃OH)≤13.996±0.005 eV. While the exact nature of the lowest-energy fragment (nominally CF₂OH⁺) is not clear, the CF⁺₃ fragment threshold leads unambiguously to ΔH°_{f 298}(CF₃OH)≥-217.2±0.9 kcal/mol and D₂₉₈(CF₃-OH)≤115.2±0.3 kcal/mol. With previously derived ΔH°_{f 298}(CF₃O)=-151.8^+1.7_-1.1 kcal/mol, this yields D₂₉₈(CF₃O-H) =117.5^+1.9_-1.4 kcal/mol, very close to, or only slightly weaker than the O-H bond energy in water: D₂₉₈(CF₃O-H)-D₂₉₈(HO-H)=-1.8 ^+1.9_-1.4 kcal/mol≈0 kcal/mol. Similarly, with the recently redetermined value for ΔH°_f(CF₂O), this implies a 298 K reaction enthalpy for the 1,2-elimination of HF from CF₃OH of 2.81^+1.7_-1.1 kcal/mol. CF₃OF and CF₃OCl have also been examined by photoionization. CF₃OF produces a very weak parent, with an apparent adiabatic IP(CF₃OF) ≤12.710±0.007 eV. An analysis of the CF⁺₃ and CF₂O⁺ fragments from CF₃OF, when combined with literature data, suggests ΔH°_{f 298}(CF₃OF)=-176.9^+1.8_-1.3 kcal/mol. The fitted value for the appearance potential of CF⁺₃ from CF₃OCl, AP₀(CF⁺₃/CF ₃OCl)≤12.85±0.01 eV, leads to ΔH°_{f 298}(CF₃OCl)≥-175.6±1.0 kcal/mol, D₂₉₈(CF₃-OCl)≤88.4±0.3 kcal/mol, and D₂₉₈(CF₃O-Cl)≤52.8^+2.0_-1.5 is kcal/mol.

Full text of DOI:10.1063/1.474017

A photoionization study of trifluoromethanol, CF 3 OH , trifluoromethyl hypofluorite, CF
3 OF , and trifluoromethyl hypochlorite, CF 3 OCl
Robert L. Asher, Evan H. Appelman, Jeffrey L. Tilson, Maritoni Litorja, Joseph Berkowitz, and Branko Ruscic
Citation: The Journal of Chemical Physics 106, 9111 (1997); doi: 10.1063/1.474017
View online: http://dx.doi.org/10.1063/1.474017
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/106/22?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Comprehensive vacuum ultraviolet photoionization study of the CF3 • trifluoromethyl radical using synchrotron
radiation
J. Chem. Phys. 136, 204304 (2012); 10.1063/1.4719529
Experimental and theoretical studies of the electronic structure and the ionization and dissociation processes of
trifluoromethyl peroxynitrate
J. Chem. Phys. 126, 184301 (2007); 10.1063/1.2734971
Combined vacuum ultraviolet laser and synchrotron pulsed field ionization study of C H 2 Br Cl
J. Chem. Phys. 126, 184304 (2007); 10.1063/1.2730829
Experimental and quantum-chemical studies on photoionization and dissociative photoionization of CH 2 Br 2
J. Chem. Phys. 120, 3270 (2004); 10.1063/1.1641010
Photoionization-efficiency spectrum and ionization energy of C 2 H 5 SCl
J. Chem. Phys. 111, 10093 (1999); 10.1063/1.480360
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 193.0.65.67
On: Tue, 23 Dec 2014 13:45:23

A photoionization study of trifluoromethanol, CF3OH, trifluoromethyl
hypofluorite, CF3OF, and trifluoromethyl hypochlorite, CF3OCl
Robert L. Asher, Evan H. Appelman, Jeffrey L. Tilson, Maritoni Litorja,
Joseph Berkowitz, and Branko Ruscic
Argonne National Laboratory, Argonne, Illinois 60439-4831
͑Received 26 November 1996; accepted 7 March 1997͒
CF3OH, an important and controversial by-product of atmospheric decomposition of CF3CFH2
͑HFC-134a͒ and other hydrofluorocarbons, has been examined by photoionization mass
spectrometry. The ionization onset is characterized by a broad Franck–Condon distribution, arising
primarily from a substantial elongation of the C–O bond upon ionization. An upper limit to the
adiabatic ionization potential ͑IP͒ of р 13.08 Ϯ 0.05 eV has been established. The appearance
potentials ͑APs͒ of the first two fragments have been accurately determined by fitting with
appropriate
model
functions
as
AP0͑CF2OHϩ/CF3OH͒р13.830Ϯ0.005 eV
and
AP0͑CFϩ3 /CF3OH͒р13.996Ϯ0.005 eV. While the exact nature of the lowest-energy fragment
͑nominally CF2OHϩ͒ is not clear, the CF3ϩ fragment threshold leads unambiguously to
⌬Hؠf 298͑CF3OH͒уϪ217.2Ϯ0.9 kcal/mol and D298͑CF3–OH͒р115.2Ϯ0.3 kcal/mol. With pre-
viously
derived
⌬Hؠf 298͑CF3O͒ϭϪ151.8Ϫϩ11..71 kcal/mol,
this
yields
D298͑CF3O–H͒
ϭ117.5ϩϪ11..49 kcal/mol, very close to, or only slightly weaker than the O–H bond energy in water:
D298͑CF3O–H͒–D298͑HO–H͒ϭϪ1.8ϩϪ11..49 kcal/molϷ0 kcal/mol. Similarly, with the recently
redetermined value for ⌬Hؠf͑CF2O͒, this implies a 298 K reaction enthalpy for the 1,2-elimination
of HF from CF3OH of 2.8ϩϪ11..17 kcal/mol. CF3OF and CF3OCl have also been examined by photo-
ionization. CF3OF produces a very weak parent, with an apparent adiabatic IP͑CF3OF͒
р12.710Ϯ0.007 eV. An analysis of the CFϩ3 and CF2Oϩ fragments from CF3OF, when combined
with literature data, suggests ⌬Hؠf 298͑CF3OF͒ϭϪ176.9ϩϪ11..83 kcal/mol. The fitted value for the
appearance potential of CFϩ3 from CF3OCl, AP0͑CF3ϩ/CF3OCl͒р12.85Ϯ0.01 eV, leads to
⌬Hؠf 298͑CF3OCl͒уϪ175.6Ϯ1.0 kcal/mol, D298͑CF3–OCl͒р88.4Ϯ0.3 kcal/mol, and D298͑CF3O–
Cl͒р52.8ϩϪ12..05 kcal/mol. © 1997 American Institute of Physics. ͓S0021-9606͑97͒03122-X͔
decomposition in the troposphere.11 Thus, it appears that the
I. INTRODUCTION
strength of the O–H bond energy is an important factor in
determining whether CF3OH is a good sink for stratospheric
CF3, or merely a long-lived reservoir.
Hydrofluorocarbons ͑HFCs͒ and hydrochlorofluorocar-
bons ͑HCFCs͒ are generally perceived as environmentally
friendlier alternatives to chlorofluorocarbons ͑CFCs͒ and ha-
lons ͑bromofluorocarbons and bromochlorofluorocarbons͒,
which have been implicated in the anthropogenic depletion
of stratospheric ozone.1 The general idea is that most of the
fluorine ends up bound as HF, thus having little effect on the
ozone budget.2 However, it has been established recently that
oxidative degradation of HFC-134a (CF3CFH2), HFC-143a
(CF3CH3), HFC-125 (CF3CF2H), HFC-23 ͑CF3H͒, HCFC-
123 (CF3CCl2H), and others can produce CF3 radicals. By
reaction with O2 or NO, these are readily converted3 to
CF3O, which may be involved in catalytic ozone removal
cycles.4 Although recent studies strongly suggest that the rel-
evant rate coefficients are low,4͑b͒,5 the ozone depletion po-
tential of HFCs ultimately depends on how efficiently
CF3O can be removed from the damaging cycle. Among
various possibilities, a prominent group of chain-termination
reactions are those that convert CF3O to CF3OH by hydrogen
abstraction from suitable donors, such as hydrocarbons6 and
perhaps even water.7,8 The ultimate fate of stratospheric
CF3OH is not clear at this time; however, it has been shown
that it has a very long lifetime with respect to unimolecular
decomposition9 ( Ͼ 104 yr) or photolysis10 ( ϳ 106 yr), with a
significantly shorter lifetime with respect to heterogeneous
From experimental observations on chlorine-initiated de-
composition of HFC-134a, Wallington et al. recently7 con-
cluded that CF3O radicals react with H2O, producing
CF3OH and OH, suggesting that the O–H bond energy in
CF3OH must be comparable to that in water, i.e.,
D298͑CF3O–H͒ϭ120Ϯ3 kcal/mol. This surprisingly high
value seems to be corroborated by a number of theoretical
studies.7,12–18 However, as Benson19 pointed out, there is a
significant volume of experimental data20 which, when com-
bined with bond additivity19,20 and other empirical
estimates,19 appears to firmly establish D298͑CF3O–H͒
ϭ109Ϯ2.5 kcal/mol ͑see Table I and also Ref. 21͒. The basic
disagreement of ϳ 10–11 kcal/mol between the ‘‘low’’ and
the ‘‘high’’ bond energy arises from the fact that the theo-
retical ⌬Hؠf(CF3OH) is lower by ϳ2–7 kcal/mol than the
empirical value, while ⌬Hؠf(CF3O) is higher by ϳ2–8
kcal/mol than the values suggested by the experiments.
In trying to resolve this issue by theoretical means,
Montgomery et al.16͑a͒ and Schneider and Wallington17 have
shown that a significant portion of the discrepancy stems
from the fact that the ab initio value ⌬Hؠf͑CF2O͒
ϭϪ145.3Ϯ1.7 kcal/mol, is ϳ 7 kcal/mol higher than the
J. Chem. Phys. 106 (22), 8 June 1997
0021-9606/97/106(22)/9111/11/$10.00
© 1997 American Institute of Physics
9111
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 193.0.65.67
On: Tue, 23 Dec 2014 13:45:23

9112
Asher et al.: Photoionization of trifluoromethyl compounds
TABLE I. Review of literature values for D298(CF3O–H), ⌬Hؠf 298(CF3OH), and ⌬Hؠf 298(CF3O) ͑all in kcal/
mol͒.
⌬Hؠf 298(CF3OH)
⌬Hؠf 298(CF3O)
Source
D298(CF3O–H)
120Ϯ3
119.4Ϯ2
•••
119.4Ϯ1.5
111.8Ϫ121.9
•••
•••
Ϫ217.7Ϯ2
Ϫ217.4
•••
•••
Wallington et al.a
Ϫ150.4Ϯ2
Schneider and Wallingtonb
Sana et al.c
•••
•••
•••
Dixon and Fernandezd
Bock et al.e
•••
Ϫ217.7Ϯ2
Ϫ217.7Ϯ2.0
Ϫ220.7
•••
Montgomery et al.f
Schneider and Wallingtong
Schneider et al.h
Ϫ149.2Ϯ2.0
Ϫ154.5
ϳ119
118.2
109
109Ϯ2.5
•••
Ϫ213.5Ϯ1.5
Ϫ215Ϯ1
Ϫ156.7Ϯ1.5
Ϫ157Ϯ1.5
Ϫ155.4Ϯ1.1
•••
Batt and Walshi
Bensonj
•••
•••
Schneider et al.k
Huey et al.l
124.7Ϯ3.6
aReference 7, based on the experimental observation that CF3O radicals react with H2O and supported by ab
initio calculations at the MP4 level.
bReference 12, theoretical value at the MP4 level.
cReference 13, theoretical value.
dReference 14, theoretical value.
eReference 15͑b͒; the range is obtained by combining theoretical values at the MP4 level from Ref. 15͑a͒ with
experimental bond energies for CH4, C2H6, and H2O.
fReference 16͑a͒, based on the theoretical value at the G2 level and isogyric reaction schemes.
gRecommended values from Ref. 17, based on a combination of various theoretical calculations and experi-
mental data from Ref. 20.
hReference 18, theoretical values at the MP2 level.
iReference 20; ⌬Hؠf(CF3O) is derived from a variety of experimental data, while ⌬Hؠf(CF3OH) is estimated
from bond additivity.
jReference 19, reiterating data from Ref. 20 and adding another estimate for ⌬Hؠf(CF3OH) based on homoth-
ermicity of OH and F.
kReference 21, reiterating experimental data from Ref. 20, but avoiding the use of JANAF’s value for
⌬Hؠf(CF3OF).
lFrom L. G. Huey, E. J. Dunlea, and C. J. Howard, J. Phys. Chem. 100, 6504 ͑1996͒; the bond energy is based
on experimental gas-phase acidity ⌬acidHؠ298͑CF3OH͒ϭ329.8Ϯ2.0 kcal/mol and an estimated electron affinity
EA͑CF3O͒ϭ108.5Ϯ3.0 kcal/mol.
colleagues26 succeeded in synthesizing this material and re-
ported that it had some measure of kinetic stability at sub-
ambient temperatures. Therefore, it appeared to us that it
might be feasible to examine this compound by photoioniza-
tion mass spectroscopy. To this end we have undertaken the
present study.
JANAF22 value, ⌬Hؠf 298͑CF2O͒ϭϪ152.7Ϯ0.4 kcal/mol, to
which the experimental19,20 ⌬Hؠf(CF3O) is referred. The
claim that the generally accepted tabulated value,22,23 which
is based on a determination of the enthalpy of hydrolysis
͑quoted to Ϯ 0.2 kcal/mol͒ and other experimental measure-
ments, is in error by such a significant amount, has been
recently investigated by us.24 Our photoionization study of
CF2O yielded ⌬Hؠf 298͑CF2O͒уϪ149.1Ϫϩ10..74 kcal/mol and in-
dicated unequivocally ͑since ours is a strict lower limit͒ that
the tabulated22,23 value is indeed too low, albeit only by 3–4
kcal/mol. However, our study also implied that the theoreti-
cal values16͑a͒,17 are most likely too high by 3–4 kcal/mol,
which brings into question the accuracy of calculated heats
of formation for other fluorine-rich compounds, including
CF3OH. Curtiss et al.25 have very recently performed a re-
assessment of GAUSSIAN-2 ͑G2͒ and density functional theo-
ries ͑DFT͒ and concluded that the largest deviations between
experiment and theory tend to occur for molecules having
multiple fluorine atoms.
II. EXPERIMENT
The photoionization apparatus employed in this study
was described recently in some detail.24 The present experi-
ments utilized primarily the He Hopfield continuum and, in a
few instances, the many-lined Werner and Lyman emission
bands of H2. The nominal photon resolution was 0.84 Å ͑full
width at half-maximum͒. All measurements of light intensity
were performed via a sodium salicylate transducer coupled to
an external photomultiplier. The H2 lines, or, in the case of
He Hopfield continuum, the superimposed Ne I, N II, and H I
lines provided an accurate internal wavelength calibration.
The uncertainty in the wavelength scale depends on the den-
sity of experimental points, and is believed to be accurate to
Ϯ 0.2 Å in most spectra presented here.
The best approach to solve this puzzle is to provide in-
dependent
experimental
evaluations
of
either
D298(CF3O–H) or ⌬Hؠf 298(CF3OH). Although CF3OH has
long been considered to belong to a category of compounds
that were too unstable to be prepared in bulk, Seppelt and his
Trifluoromethanol, CF3OH, was prepared by the reac-
tion of trifluoromethyl hypochlorite, CF3OCl, with HCl in
J. Chem. Phys., Vol. 106, No. 22, 8 June 1997
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 193.0.65.67
On: Tue, 23 Dec 2014 13:45:23

Asher et al.: Photoionization of trifluoromethyl compounds
9113
III. COMPUTATIONAL DETAILS
CClF3 solvent as the mixture was gradually warmed from
liquid-nitrogen temperature to Ϫ 100 °C.26͑b͒ The reaction
was carried out in a 3/4 in.-o.d. Kel-F tube fitted with a
Kel-F needle valve. The solvent and excess HCl were re-
moved by pumping at Ϫ 130 °C, and the Cl2 by-product at
Ϫ 100 °C. Trifluoromethanol was stored at liquid-nitrogen
temperature in the reaction vessel. No decomposition was
observed even after many days of storage.
The supporting ab initio calculations on CF3OH were
performed using the GAUSSIAN 94 package.28 Geometries of
neutral and ionic species were optimized at the MP2
͑second-order Mo”ller–Plesset͒ level using the 6-31 G(d,p)
basis set.29 Prior to use, the calculated harmonic frequencies
were scaled by 0.89.
The thermodynamically relevant experimental fragmen-
tation thresholds were obtained by least-squares fits of the
data with appropriate model functions. This approach has
been shown to produce appearance potentials of significantly
greater accuracy and reliability than graphical extrapolation
methods.24,30,31 The relevant details of this method have been
presented recently elsewhere.24 The model functions used
here have been obtained by convoluting a flexible kernel of
Trifluoromethyl hypochlorite, CF3OCl, and trifluorom-
ethyl hypofluorite, CF3OF, were prepared by the cesium-
fluoride-catalyzed addition of ClF and F2, respectively, to
carbonyl fluoride, CF2O, at temperatures between Ϫ 78 °C
and ambient.27 For both preparations, the anhydrous CsF
catalyst was dried before use by heating in an atmosphere of
F2. The catalyst was mixed with 3/16-in.-diam stainless steel
balls in a Monel reaction vessel to permit effective agitation.
The product was distilled from the reaction vessel into a
Monel pressure vessel, in which it was stored at ambient
temperature. No decomposition of either compound was ob-
served during storage.
Technical-grade CF2O was obtained from PCR, Inc. and
ClF from Ozark–Mahoning. Chlorotrifluoromethane,
CClF3, and anhydrous HCl were obtained from AGA Spe-
cialty Gases, and anhydrous cesium fluoride was purchased
from Aldrich.
The flow of CF3OH into the photoionization apparatus
was controlled by regulating the temperature of the sample
vessel with a suitable slush bath, rather than by throttling
with a valve. The best results were obtained with a slush of
‘‘aged’’ methanol, which had picked up some moisture from
the air and, depending on its composition, could hold a stable
temperature between Ϫ 98 and Ϫ 104 °C throughout the day.
During our initial attempt to admit the CF3OH into the in-
strument, we were only able to detect the decomposition
products CF2O and HF. By making successive modifications
to the inlet system, we were able to improve significantly the
ratio of CF3OH to its decomposition products. The changes
consisted of shortening as much as possible the transport
path from the sample vessel to the ionization region, as well
as removing all metal surfaces from the sample path and
replacing them with components fabricated from Teflon™ or
Kel-F. In the process we learned that even the smallest and
seemingly unimportant metal components ͑such as a 2-cm-
long piece of 1/4-in.-o.d. tubing inside the vacuum, at the
inlet of the ionization cell͒ can cause considerable decompo-
sition of CF3OH.
Both the trifluoromethylhypochlorite, CF3OCl, and the
trifluoromethylhypofluorite, CF3OF, were admitted directly
into the instrument from their Monel storage cylinders. The
sample pressure was reduced by cooling the cylinder in an
appropriate slush bath; acetone/dry-ice (Ϫ78 °C) was used
for CF3OCl; and n-pentane (Ϫ131 °C) for CF3OF. The
sample flow was further controlled by a relatively coarse
inlet needle valve. Since CF3OCl was causing a very rapid
deterioration of the particle multiplier, the number of mea-
surements performed with this sample was kept to a bare
minimum.
the form 1 Ϫ exp Ϫ B(h␯ Ϫ E ) with an internal energy dis-
͓
͔
͖
͕
T
␩
tribution function of the form E exp( Ϫ aE), where h␯ is the
photon energy, ET is the fragmentation threshold, and B, ␩,
and a are adjustable parameters.24 Parameter ␩ of the func-
tion representing the internal energy distribution of the neu-
tral parent was predetermined with the aid of Haarhoff’s
expression32 for the density of states, computed numerically
in the range of interest by using either calculated
(CF3OH), experimental33 (CF3OF), or estimated (CF3OCl)
frequencies.34 Parameter a was obtained by imposing the
requirement that the overall function corresponds to the cor-
rect amount of average internal energy available for frag-
mentation at 298 K ͑0.090 eV for CF3OH, 0.113 eV for
CF3OF, and ϳ 0.12 eV for CF3OCl͒. In most cases where the
model function provides an adequate description of the ob-
served threshold, and the statistical scatter of the experimen-
tal points is small or moderate, the quality of the fit ͑as
determined by the Gaussian criterion͒ is a surprisingly sen-
sitive function of the threshold value ET , with an associated
uncertainty similar to, and often even smaller, than the un-
certainty in the wavelength calibration.
IV. RESULTS
A. Overview and parent ionization of CF3OH
Table II lists a typical mass spectrum of CF3OH ob-
served at the He I resonance line at 584.33 Å ͑21.218 eV͒.
The intensities are not corrected for the inherent mass dis-
crimination function of the quadrupole mass spectrometer. A
fraction of m/e ϭ 66 (CF2Oϩ), and essentially all of m/e
ϭ 20 (HFϩ) can be attributed to ionization of CF2O and HF,
since some CF3OH was heterogeneously decomposing in the
inlet in spite of all precautions. Using the ratio of signals24 at
m/e ϭ 66 and 47 from pure CF2O, one concludes that no
more than 1/3 of m/e ϭ 47 (FCOϩ) in Table II can be attrib-
uted to fragmentation of CF2O. Therefore, CFϩ3 and FCOϩ
are the two most prominent fragments from CF3OH, fol-
lowed by CF2OHϩ. Parent ionization contributes less than
10% to the total ionization.
The He I mass spectrum in Table II shows that m/e
ϭ 85 ͑CF3Oϩ fragment͒ is at the lower limit of detectability.
This frustrates the most direct approach to determine
J. Chem. Phys., Vol. 106, No. 22, 8 June 1997
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 193.0.65.67
On: Tue, 23 Dec 2014 13:45:23

9114
Asher et al.: Photoionization of trifluoromethyl compounds
TABLE II. A typical photoionization mass spectrum of the CF3OH sample
at 584.33 Å ϵ 21.218 eV. The intensities are not corrected for the quadru-
pole mass discrimination function.
Species
Relative intensity
m/e
86
85
69
67
66
50
47
20
CF3OHϩ
CF3Oϩ
CFϩ3
11.5
Ͻ0.05
91.1
CF2OHϩ
CF2Oϩ
CFϩ2
39.6
͓10.9͔
a
3.7b
100.0c
6.1d
FCOϩ
HFϩ
aHeavily contaminated by CF2O from heterogeneous unimolecular decom-
position of CF3OH.
bUp to 1/4 of this intensity may be due to fragmentation from the decom-
position product CF2O.
cNot more than 1/3 of this intensity is due to fragmentation of the decom-
position product CF2O.
dMostly from heterogeneous unimolecular decomposition of CF3OH.
FIG. 1. An overview of the photoion yield curves of the parent CF3OHϩ and
its principal fragments, covering ϳ 950–650 Å. At shorter wavelength only
sparse points have been recorded. The relative abundance of all species is
correctly depicted in the figure, apart from quadrupole transmission factors.
The relatively small signals corresponding to CF2Oϩ and HFϩ ͑ϳ 1/12 and
ϳ 1/16 of the CFϩ3 intensity at ϳ 750 Å, respectively͒ were omitted, since
they were attributable to parent ionization of the heterogeneous decomposi-
tion products ͑CF3OH→CF2OϩHF͒. The FCOϩ signal shown is corrected
for the contribution arising from fragmentation of CF2O.
D0(CF3O–H), which would have been to determine the ap-
pearance potential ͑AP͒ of the CF3Oϩ fragment from
CF3OH, and to combine it with a separately determined ion-
ization potential ͑IP͒ of CF3O, i.e., D0͑CF3O–H͒
ϭAP0͑CF3Oϩ/CF3OH͒ϪIP͑CF3O͒. This approach would
also have to account for the possibility that the m/e ϭ 85
fragment can correspond to the isomeric ͑and likely more
stable͒ CF2OFϩ. The fact that fragmentation to CF2OFϩ re-
quires rearrangement, while CF3Oϩ may be less favored en-
ergetically, is probably the inherent reason for the weakness
of m/e ϭ 85 fragment.
The next best approach to obtain D(CF3O–H) is to de-
termine ⌬Hؠf(CF3OH) from the appearance potential of a
suitable fragment. Table II suggests that CFϩ3 is probably the
best candidate for this purpose. Of the other fragments, the
heat of formation of CF2OHϩ is less well established, while
CF2Oϩ and FCOϩ are very difficult to use because of con-
comitant signals from the decomposition product CF2O,
which occur at slightly lower energy. Furthermore, FCOϩ
from CF3OH corresponds to a second-generation fragment,
and is not a prime candidate for thermodynamical inferences.
Figure 1 shows an overview of the photoion yield sig-
nals obtained from CF3OH between ϳ 950 and 650 Å. The
relative intensities of the ion yield curves reflect the true
abundances as measured through our quadrupole mass filter.
The FCOϩ signal is corrected for the contribution arising
from fragmentation of CF2O, while the relatively small sig-
nals corresponding to CF2Oϩ and HFϩ ͑ϳ 1/12 and ϳ 1/16 of
the CFϩ3 intensity at ϳ 750 Å, respectively͒ are omitted alto-
gether. They were interpreted as arising predominantly from
parent ionization of the heterogeneous decomposition prod-
ucts, because their shapes matched rather closely known
spectra of CF2O and HF.
CFϩ3 appear relatively early, at ϳ 900 Å. Figure 2 displays
the threshold regions of CF3OHϩ, CF2OHϩ, and CF3ϩ in
greater detail. From its onset, which appears to be rather
rounded, the CF3OHϩ parent ion keeps growing toward
shorter wavelength until it culminates in a cusp ( ϳ 902 Å),
at which point the CF2OHϩ fragment starts appearing. The
FIG. 2. The threshold regions of CF3OHϩ, CF2OHϩ, and CF3ϩ from
CF3OH. The parent ion displays a rounded onset, which is a consequence of
a significant change in geometry upon ionization. The adiabatic threshold
can be muddled by the presence of hot bands, which may be difficult to
distinguish from vanishingly small Franck–Condon factors. Toward shorter
wavelength, the growth of CF3OHϩ terminates in a cusp coinciding with the
onset of the first fragment, CF2OHϩ, which experiences a similar effect at
the onset of the next fragment, CFϩ3 .
Figure 1 shows that parent ionization of CF3OH corre-
sponds to a relatively small overall contribution. The
CF3OHϩ yield attains its maximum at ϳ 900 Å and, after a
slight decline, maintains essentially a constant intensity as
one proceeds to shorter wavelengths. Both CF2OHϩ and
J. Chem. Phys., Vol. 106, No. 22, 8 June 1997
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 193.0.65.67
On: Tue, 23 Dec 2014 13:45:23

Asher et al.: Photoionization of trifluoromethyl compounds
9115
TABLE III. MP2/6-31G(d,p) optimized geometries of CF3OH and
CF3OHϩ and scaled harmonic vibrational frequencies.a
CF3OHϩ
CF3OH
1.351
1.332
1.352
0.966
1.651
1.279
1.290
0.997
102.2
102.3
112.0
180.0
r(CO)/Å
r(CF )/Å
Ј
r(CF)/Å
r(OH)/Å
ЄF CO/°
108.3
Ј
ЄFCO/°
ЄCOH/°
112.2
108.1
180.0
ЄF COH/°
Ј
Frequencies/cmϪ1
3468
1302
1200
1116
1026
810
564
555
535
403
3146
1366
907
1311
842
664
354
504
503
286
291
FIG. 3. The threshold region of the CFϩ3 fragment from CF3OH. The solid
line is a least-squares fit with a threshold model function ͑see the text͒
yielding AP0͑CFϩ3 /CF3OH͒р13.996Ϯ0.005 eV.
391
͓208͔
b
b
͓63͔
aCalculated harmonic frequencies are scaled by a factor of 0.89.
bFrequency corresponding to internal rotation.
lated changes in geometry, and assuming that the normal
modes in the ion correspond to orthogonal harmonic
oscillators,35 we constructed a simulated Franck–Condon en-
velope. Surprisingly, the influence of room-temperature hot
bands was relatively minor, and appeared to shift the appar-
ent threshold toward lower energy only by ϳ 0.04 eV.
Therefore, correcting the experimental onset by a similar
growth of CF2OHϩ is, in turn, terminated in a cusp related to
the onset of the CFϩ3 fragment. Such a pattern, where growth
is stopped rather suddenly at the onset of the subsequent
fragmentation process, can be interpreted as a sign that the
internal energy of the parent is well randomized and that
quasi equilibrium theory provides an adequate description of
the unimolecular decomposition. Thus, the fragmentation on-
sets of both CF2OHϩ and CF3ϩ look promising in providing
thermodynamically useful information.
The pronouncedly rounded onset of the parent is an in-
dication of a broad Franck–Condon envelope, reflecting a
significant change in geometry upon ionization. Upon close
scrutiny of several independent scans, it was determined that
the first detectable departure of the parent ion signal from the
background level occurs at 951 Ϯ 2 Å ϵ 13.04 Ϯ 0.03 eV.
This would nominally correspond to an upper limit to
IP͑CF3OH͒, if one were to disregard possible obfuscation
from hot bands.
The importance of hot bands and the nature of the
change in geometry upon ionization was further investigated
by theoretical methods. The predicted adiabatic IP of
12.65 eV Ϸ 980 Å ͑at the MP2 level and including zero-point
energy corrections͒ is much lower than the experimentally
observed onset. The vertical IP is predicted to occur
ϳ 0.95 eV higher (13.60 eV Ϸ 910 Å), implying a very
broad Franck– Condon envelope. Table III lists MP2 opti-
mized geometries and scaled frequencies for CF3OH and its
cation. The values for the neutral species are similar to those
published previously.16 The calculated geometries indicate
that ionization causes a considerable increase ͑0.300 Å͒ in
the C–O bond length, and is expected to cause extended
progressions of ϳ 660 and 350 cmϪ1. Based on the calcu-
amount yields
a
conservative upper limit of
IP͑CF3OH͒р13.08Ϯ0.05 eV.
B. The CF؉3 and CF2OH؉ fragments from CF3OH
Figure 3 displays the threshold region of the CFϩ3 frag-
ment in more detail. The solid line is a least-squares fit
with a threshold model function, obtained as outlined in Sec.
III. The fitted threshold value is AP298͑CFϩ3 /CF3OH͒
р13.906Ϯ0.005 eV ͑р 13.996 Ϯ 0.005 eV at 0 K͒.
Figure 4 shows the threshold region of the CF2OHϩ
fragment in more detail. As opposed to the threshold region
of the CFϩ3 fragment, CF2OHϩ appears to display a weak
sloping tail, extending to lower energy. The fit ͑the solid line
in Fig. 4͒ accommodates for this by incorporating a linear
background, with the relevant parameters obtained by prefit-
ting the tail region between 13.2 and 13.4 eV. Apart from
this inclusion of a linear background, the model curve is
analogous to that for CFϩ3 , and uses the identical internal
energy function. The resulting threshold value is
AP298͑CF2OHϩ/CF3OH͒р13.740Ϯ0.005 eV ͑р13.830Ϯ0.005
eV at 0 K͒. These, as well as all other thermodynamically
relevant values obtained in this work are summarized in
Table IV.
C. Photoionization of CF3OF
Figure 5 provides an overview of the ion yield curves
resulting from photoionization of CF3OF. The relative ion
yields reflect the intensities as measured through our quad-
J. Chem. Phys., Vol. 106, No. 22, 8 June 1997
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 193.0.65.67
On: Tue, 23 Dec 2014 13:45:23

9116
Asher et al.: Photoionization of trifluoromethyl compounds
FIG. 4. The threshold region of the CF2OHϩ fragment from CF3OH. The
solid line is a least-squares fit with a threshold model function ͑see the text͒
yielding AP0͑CF2OHϩ/CF3OH͒р13.830Ϯ0.005 eV.
FIG. 5. An overview of the photoion yield curves of the parent CF3OFϩ and
its principal fragments, covering ϳ 1000–700 Å. The relative abundances
of various ions are correctly depicted in the figure, apart from quadrupole
transmission factors. The parent CF3OFϩ is very weak, but measurable.
Discernible in the CF2Oϩ signal as two mild humps ͑at ϳ 800 and
ϳ 845 Å͒ is a small contribution from parent ionization of the CF2O precur-
sor used in the synthesis of CF3OF and present as a minor impurity. The
FCOϩ curve is omitted from this figure, since it was completely attributable
to fragmentation from this CF2O impurity.
rupole mass filter. The most prominent species is the CF3ϩ
fragment. The CF2Oϩ curve contains a small contribution
from parent ionization of the residual CF2O precursor used in
the synthesis of CF3OF. The FCOϩ curve was omitted from
Fig. 5, because it was completely attributable to fragmenta-
tion from the CF2O impurity.
weakness of the parent signal, one concludes that the adia-
batic ionization onset of CF3OF is Ͻ 12.782 Ϯ 0.007 eV
(970.0 Ϯ 0.5 Å), and most likely р 12.710 Ϯ 0.007 eV
(975.5 Ϯ 0.5 Å). This is significantly lower than the value of
13.0 eV listed in the compilation of Lias et al.36 A cusplike
feature, centered at 956 Ϯ 1 Å (12.97 Ϯ 0.01 eV), denotes the
The parent CF3OFϩ ion is very weak, but measurable.
Immediately following the ionization onset it shows a very
brief region of growth, after which it essentially levels off.
The threshold region is depicted in more detail in Fig. 6, with
points at 0.5 Å intervals. With some prudence due to the
TABLE IV. Summary of measured and deduced thermodynamically relevant quantities.
Quantity
IP͑CF3OH͒
298 K
0 K
•••
р13.08Ϯ0.05 eV
р13.996Ϯ0.005 eV
р13.830Ϯ0.005 eV
р12.710Ϯ0.007 eV
(Ͻ12.782Ϯ0.007 eV)
ϳ13.0Ϯ0.1 eV
(Ͻ13.38Ϯ0.10 eV)
ϳ12.97Ϯ0.01 eV
ϳ14.2Ϯ0.1 eV
AP͑CFϩ3 /CF3OH͒
AP͑CF2OHϩ/CF3OH)
IP͑CF3OF͒
р13.906Ϯ0.005 eV
р13.740Ϯ0.005 eV
•••
AP͑CFϩ3 /CF3OF͒
ϳ12.9Ϯ0.1 eV
(Ͻ13.27Ϯ0.10 eV)
•••
ϳ14.1Ϯ0.1 eV
р12.736Ϯ0.007 eV
AP͑CF2Oϩ/CF3OF͒
AP͑CFϩ3 /CF3OCl͒
⌬Hؠf(CF3OH)
р12.85Ϯ0.01 eV
Ϫ217.2Ϯ0.9 kcal mol
Ϫ215.3Ϯ0.9 kcal/mol
⌬Hؠf(CF3OF)
Ϫ176.9
kcal/mol
Ϫ175.5
kcal/mol
Ϫ1.3
ϩ1.8
Ϫ1.3
ϩ1.8
(ϾϪ186.0Ϯ3.4 kcal/mol)
(ϾϪ184.6Ϯ3.4 kcal/mol)
⌬Hؠf(CF3OCl)
Ϫ175.6Ϯ1.0 kcal/mol
Ϫ174.2Ϯ0.9 kcal/mol
ϩ1.7
Ϫ1.1
ϩ1.7
⌬Hؠr(CF3OH→CF2OϩHF)
⌬Hؠf(CF2OHϩ)
PA͑CF2O͒
2.8
kcal/mol
1.6
kcal/mol
Ϫ1.1
р84.2Ϯ1.0 kcal/mol
р85.2Ϯ0.9 kcal/mol
ϩ1.7
•••
у132.4
117.5
kcal/mol
Ϫ1.2
ϩ1.9
ϩ1.9
D(CF3O–H)
kcal/mol
116.4
kcal/mol
Ϫ1.4
Ϫ1.4
D(CF3–OH)
D(CF3O–Cl)
115.2Ϯ0.3 kcal/mol
113.9Ϯ0.3 kcal/mol
ϩ2.0
ϩ2.0
52.8
kcal/mol
52.3
kcal/mol
Ϫ1.5
Ϫ1.5
D(CF3–OCl)
D(CF3O–H)–D(HO–H)
88.4Ϯ0.3 kcal/mol
87.6Ϯ0.3 kcal/mol
ϩ1.9
ϩ1.9
Ϫ1.8
kcal/mol
Ϫ1.7
kcal/mol
Ϫ1.4
Ϫ1.4
J. Chem. Phys., Vol. 106, No. 22, 8 June 1997
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 193.0.65.67
On: Tue, 23 Dec 2014 13:45:23

Asher et al.: Photoionization of trifluoromethyl compounds
9117
FIG. 6. The threshold region of the parent CF3OFϩ ion. The growth of this
very weak species is terminated in a cusplike structure (956 Ϯ 1 Å) corre-
sponding to the onset of the first fragment, CFϩ3 .
FIG. 8. The shape of the CF2Oϩ fragment yield curve from CF3OF after
subtracting the contribution from the CF2O impurity. The solid line is an
approximate fit ͑see the text͒, yielding AP0͑CF2Oϩ/CF3OF͒Ϸ14.2Ϯ0.1 eV.
termination of the growth of the parent signal and the onset
of the CFϩ3 fragment.
Figure 7 depicts the threshold region of the CFϩ3 frag-
ment in greater detail. It should be noted that the fragment
curve displays a long tail extending below the measurable
ionization onset of the parent. Although unusual, this is not
entirely surprising, since the internal energy of the neutral
CF3OF should, at least in principle, be available for fragmen-
tation even if the nominal photon energy is below the adia-
batic IP. Somewhat more surprising is the appearance of a
small humplike structure in the tail region, centered almost
exactly at the ionization onset of CF3OF, ϳ 12.70 eV. The
origin of this feature is not clearly understood at this time,
and perhaps relates to some second-order process.
described in Sec. IV B.37 Clearly, the fit is far from perfect,
and it misses most of the curvature in the tail region. Thus,
the fitted threshold value of 13.38 Ϯ 0.10 eV ͑at 0 K͒ is just
an upper limit to the true value. In all likelihood, the frag-
mentation threshold is distorted by Frank–Condon factors
for parent ionization, as was recently31 found to be the case
in CF3Cl. This inference is supported by the very small gap
between the onsets of the CF3OFϩ parent and CF3ϩ fragment,
and the related weakness of the parent, both of which are
even more pronounced than in CF3Cl.
Given these complications, it appears that a better esti-
mate of the CFϩ3 fragmentation onset can be derived from the
position of the cusplike termination of growth of the parent
͑Fig. 6͒, found to occur at 12.97 Ϯ 0.01 eV. Another estimate
can be obtained from examining the amount of misfit in Fig.
8; this leads to ϳ 13.0 Ϯ 0.1 eV. The CF2Oϩ fragment can
provide an additional handle on the thermochemistry of
CF3OF. Figure 8 displays the shape of this fragment yield
curve, after subtracting the contribution attributable to parent
ionization of the CF2O impurity present in the sample. The
amount of correction was determined from the requirement
that two broad peaklike structures, visible in Fig. 5 at
ϳ 845 Å ( ϳ 14.7 eV) and ϳ 800 ( ϳ 15.5 eV) and originat-
ing from direct ionization of CF2O, disappear after subtrac-
tion of the known24 parent spectrum. Separately, the related
amount of the FCOϩ fragment attributable to ionization of
CF2O was used to reduce the measured FCOϩ signal from
the CF3OF sample; the level of correction for the CF2O im-
purity that optimally compensated the humps in the CF2Oϩ
ion yield also completely ͑and rather exactly͒ canceled the
FCOϩ signal recorded from the CF3OF sample. The cor-
rected CF2Oϩ fragment yield in Fig. 8 displays a low-
intensity pseudolinear fragmentation onset, followed by a
second, much more pronounced growth at higher energy,
which correlates with the termination of growth of the
CF2OFϩ fragment and presumably corresponds to the sec-
ondary fragmentation to CF2Oϩϩ2F. The weakness of the
primary fragmentation process leading to CF2OϩϩF2, is at-
tributable to steric hindrance, since it requires the concerted
The solid line in Fig. 7 is an attempted model fit of the
experimental data, carried out in a fashion analogous to that
FIG. 7. The threshold region of the CFϩ3 fragment from CF3OF. The frag-
ment displays a long tail extending below the nominal ionization onset of
the parent, and an unusual humplike structure centered at the adiabatic IP of
the parent. The solid line is an attempted fit with a threshold model function
͑see the text͒ and provides only an upper limit to the AP. The model is
unable to reproduce the roundness in the tail, mainly because the fragment
yield curve is distorted by unfavorable Franck–Condon factors for parent
ionization.
J. Chem. Phys., Vol. 106, No. 22, 8 June 1997
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 193.0.65.67
On: Tue, 23 Dec 2014 13:45:23

9118
Asher et al.: Photoionization of trifluoromethyl compounds
уϪ215.3Ϯ0.9 kcal/mol, equivalent to ⌬Hؠf 298͑CF3OH͒
уϪ217.2Ϯ0.9 kcal/mol. This value is extremely close to
several theoretical estimates,12,13,16,17 but discernibly lower
than the estimates based on group additivity properties.19,20
Subtracting the best currently available value31 of
IP͑CF3͒ϭ9.055Ϯ0.011 eV from our AP͑CFϩ3 /CF3OH͒ di-
rectly yields D0͑CF3–OH͒р113.9Ϯ0.3 kcal/mol ͑р 115.2
Ϯ 0.3 kcal/mol at 298 K͒. The path to D(CF3–OH) is
more circuitous and requires the knowledge of
⌬Hؠf(CF3O).
We
have
recently
recommended24
⌬Hؠf 298͑CF3O͒ϭϪ151.8Ϫϩ11..77 kcal/mol, based on our new
value24 for ⌬Hؠf 298͑CF2O͒ϭ Ϫ149.1Ϫϩ10..47 kcal/mol and the
room-temperature enthalpy of 21.7 Ϯ 0.9 kcal/mol for the re-
action CF3O→CF2OϩF that can be derived from data listed
by Batt and Walsh.20 The latter reaction enthalpy does not
appear to be contested, since it is very close to 22.9
Ϯ 2.6 kcal/mol, which is implied by the theoretical12,17
⌬Hؠf values for CF2O and CF3O.
FIG. 9. The threshold region of the CFϩ3 fragment from CF3OCl. The solid
line is a least-squares fit with a threshold model function ͑see the text͒,
yielding AP0͑CF3ϩ/CF3OCl͒р12.85Ϯ0.01 eV.
With ⌬Hؠf(CF3OH) derived above, the recommended
value for ⌬Hؠf(CF3O)
leads to D298͑CF3O–H͒
ϩ1.9
ϭ117.5ϩϪ11..94 kcal/mol ͑or 116.4
kcal/mol at 0 K͒, imply-
Ϫ1.4
ing that the O–H bond energy in CF3OH is more than 13
kcal/mol higher than the analogous bond energy in
methanol38 and quite close to the O–H bond in water.23
In fact, when the error bar is considered,
elimination of fluorine atoms from opposite ends of the
CF3OF molecule. In addition, the process has to compete
with the well-developed fragmentation generating CFϩ3 . The
solid line in Fig. 8 is an approximate fit by a model function,
which incorporates two linear kernels and ‘‘fluctuations’’ for
the higher energy secondary fragmentation.24 While the ap-
proximate nature of the fit makes the onset of the secondary
process highly uncertain, the fit suggests a tentative appear-
ance potential for the primary process of 14.2 Ϯ 0.1 eV ͑0 K͒.
D
298(CF3O–H)ϪD298͑HO–H͒ϭϪ1.8ϩϪ11..49 kcal/molϷ0 kcal/
mol. The present result is significantly closer to the high
bond energy of ϳ 120 kcal/mol proposed originally by Wall-
ington et al.7 than to the low value of ϳ 109 kcal/mol sug-
gested by Benson.19 The relatively high bond energy sug-
gests that CF3OH is a reasonably good sink for stratospheric
CF3O. However, since it seems that nominally D(HO–H)
у D(CF3O–H), reaction with OH radicals may be capable
of recycling CF3O from CF3OH.
D. The CF؉3 fragment from CF3OCl
The value of ⌬Hؠf(CF3OH) derived above also implies
that at 298 K the reaction enthalpy for the 1,2-elimination of
With CF3OCl we focused our attention primarily on the
CFϩ3 fragment ion yield curve, given the deleterious effect
that this sample had on our particle multipliers. The parent
ionization of CF3OCl and other fragments were explored
only in a very cursory fashion, and no attempt was made to
construct an overview spectrum. Nevertheless, it was pos-
sible to establish clearly that parent ionization constitutes a
significant portion of the total ionization yield, and that
CFϩ3 is an early and strong fragment. For example, at the
Ne I line (743.7195 Å ϵ 16.671 eV͒, CFϩ3 appeared to be the
most abundant species, while the parent CF3O35Cl was the
next largest signal at about half of that intensity.
ϩ1.7
Ϫ1.1
ϩ1.7
kcal/mol at 0 K͒. This, as
Ϫ1.1
HF is 2.8
kcal/mol ͑1.6
well as other thermodynamical values derived in this work
are summarized in Table IV.
B. ⌬Hؠf(CF2OH؉)
Combining ⌬Hؠf(CF3OH) determined above and
AP͑CF2OHϩ/CF3OH͒ leads to ⌬Hؠ ͑CF2OHϩ͒р85.2
f 0
Ϯ0.9 kcal/mol ͑or 84.2 Ϯ 1.0 kcal/mol at 298 K͒. This is
ϳ 32 kcal/mol higher than the heat of formation ͑52 kcal/
mol͒ listed by Lias et al.36 and derived from the generally
accepted value for the proton affinity39 PA͑CF2O͒
ϭ160.5 kcal/mol ͑based on measurements by McMahon and
collaborators40͒ and the JANAF22 value for ⌬Hؠf(CF2O). Us-
ing the new24 ⌬Hؠf(CF2O) instead of the tabulated value, we
obtain 56.1 Ϯ 2 kcal/mol ͑298 K͒ for the heat of formation of
protonated carbonyl fluoride, still 28.1 kcal/mol lower than
the value implied by AP͑CF2OHϩ/CF3OH͒.
Figure 9 displays the threshold region of CFϩ3 from
CF3OCl. The solid line is the model fit, carried out as de-
scribed above for CF3OH and CF3OF. The fitted threshold is
AP298͑CFϩ3 /CF3OCl͒р12.736Ϯ0.007 eV, corresponding to
12.85 Ϯ 0.01 eV at 0 K.
V. DISCUSSION
A. ⌬Hؠf(CF3OH), D0(CF3–OH), and D0(CF3O–H)
Technically, experimental fragmentation thresholds are
upper limits to the true values, and therefore the discrepancy
is in the allowed direction. However, the shape of the ion
yield curve and the quality of the threshold fit in Fig. 4
We have recently established31 that ⌬Hؠf 0͑CF3ϩ͒
ϭ98.1Ϯ0.9 kcal/mol. With23 ⌬Hؠf 0͑OH͒ϭ9.35Ϯ0.05 kcal/
mol, our AP͑CFϩ3 /CF3OH͒ leads to ⌬Hؠf 0͑CF3OH͒
J. Chem. Phys., Vol. 106, No. 22, 8 June 1997
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 193.0.65.67
On: Tue, 23 Dec 2014 13:45:23

Asher et al.: Photoionization of trifluoromethyl compounds
9119
AP0͑CF2Oϩ/CF3OF͒Ϸ14.2Ϯ0.1 eV,
when
suggest that our value is more than just a coarse upper limit.
proximate
coupled with24 IP͑CF2O͒ϭ13.024Ϯ0.004 eV, yields 28
Ϯ 2 kcal/mol, further corroborating the reaction enthalpy
derived from Batt and Walsh.20 This yields
The only hint for a possible existence of a lower energy
fragmentation process is the weak sloping tail extending to
lower energy. The shape of the threshold suggests that the
fragmentation either does not involve rearrangement, or that
such rearrangement is unusually facile and efficient. In the
absence of rearrangement, the isomeric structure most likely
generated by the fragmentation of CF3OHϩ is CF2OHϩ, cor-
responding to a simple bond scission.
The massive discrepancy between the values of
⌬Hؠf(CF2OHϩ), as obtained from AP͑CF2OHϩ/CF3OH) or
PA͑CF2O) is puzzling. The heat of formation of CF2OHϩ
implied by PA͑CF2O) is such that fragmentation of CF3OH
would have to commence at ϳ 12.6 eV, below the observ-
able ionization threshold. Consequently, one would expect
the parent to be very weak and the primary process at the
adiabatic ionization onset to be the appearance of the
CF2OHϩ fragment. This is clearly not the case here. One
possible explanation is that the CF2OHϩ asymptote is not
readily reachable from the CF3OHϩ manifold accessible by
direct ionization, perhaps because of a barrier. In this case
the measured onset of CF2OHϩ would correspond either to
the top of the barrier or to some suitable excited state of
CF2OHϩ, which effectively opens a new dissociation asymp-
tote. A slight variation on this theme is that the observed
fragmentation processes and the PA experiments access dif-
ferent isomers of protonated carbonyl fluoride.
⌬Hؠf 298͑CF3OF͒ϭϪ176.9ϩϪ11..83 kcal/mol, or Ϫ 175.5ϩ1.8
Ϫ1.3 kcal/
mol at 0 K, over 5 kcal/mol higher than the JANAF22 value.
Using this heat of formation of CF3OF, one obtains
AP0͑CFϩ3 /CF3OF͒ϭ12.99Ϯ0.13 eV, significantly lower than
our fitted upper limit of 13.4 Ϯ 0.1 eV, but very close to the
estimates based either on the amount of misfit in Fig. 7 or on
the position of the cusplike structure in the parent CF3OFϩ
ion yield ͑Fig. 6͒.
D. ⌬Hؠf(CF3OCl)
Our AP0͑CFϩ3 /CF3OCl͒, with the well established22,23
⌬Hؠf 0͑OCl͒ϭ24.15Ϯ0.03 kcal/mol yields ⌬Hؠf 0͑CF3OCl͒
уϪ174.2Ϯ0.9 kcal/mol, equivalent to Ϫ175.6Ϯ1.0
kcal/mol at 298 K. This in turn leads to
D0͑CF3O–Cl͒р52.3ϩϪ12..50 kcal/mol ͑52.8
kcal/mol at 298
ϩ2.0
Ϫ1.5
K͒, which compares very well with the upper limit of
Ͻ 53 kcal/mol determined experimentally by Rengarajan
et al.43 The value of AP0͑CF3ϩ/CF3OCl͒, when combined
with IP͑CF3͒, produces D0͑CF3–OCl͒р87.6Ϯ0.3 kcal/mol,
equivalent to 88.4 Ϯ 0.3 kcal/mol at 298 K.
VI. CONCLUSIONS
In order to shed some more light on this situation, we
have performed exploratory ab initio calculations at the
MP2/6-31G(d,p) level. These were unable to indicate any
singlet or triplet structures that would be lower in energy
than CF2OHϩ. Instead, a structure corresponding to FCOϩ
with an HF moiety roughly perpendicular to it and ϳ 2.2 Å
away ͑C–F distance͒, was found to represent a minimum on
the potential energy surface roughly midway between the
most stable CF2OHϩ structure and the FCOϩϩHF asymp-
tote. However, even if the observed threshold corresponds to
such an adduct, it is not apparent why the fragmentation
pathways of CF3OHϩ would ignore the generation of
CF2OHϩ by a simple bond scission. Clearly, a more elabo-
rate theoretical study of the CF3OHϩ and CF2OHϩ systems
is needed before a definitive answer can be given.
Our study of CF3OH by photoionization mass spectrom-
etry yields IP͑CF3OH͒р13.08Ϯ0.05 eV, and, by accurate
fitting of the fragmentation threshold regions, AP0͑CFϩ3 /
CF3OH͒р13.996Ϯ0.005 eV and AP0͑CF2OHϩ/CF3OH͒
р13.830Ϯ0.005 eV. The appearance potential of the CF3ϩ
fragment from CF3OH proves useful in deriving
⌬Hؠf 298͑CF3OH͒уϪ217.2Ϯ0.9 kcal/mol, corroborating re-
cently calculated ab initio values12,13,16,17 for this quantity.
Together with IP͑CF3͒ϭ9.055Ϯ0.011 eV, it also implies
D298͑CF3–OH͒р115.2Ϯ0.3 kcal/mol. Using the recently24
recommended ⌬Hؠf(CF2O) the present value of
⌬Hؠf(CF3OH) leads to a reaction enthalpy for the 1,2-
ϩ1.7
Ϫ1.1
ϩ1.7
kcal/mol ͑1.6
Ϫ1.1
elimination of HF of ⌬Hؠr 298ϭ2.8
kcal/mol at 0 K͒.
Combining ⌬Hؠf(CF3OH) with ⌬Hؠf(CF3O) derived
previously,24 leads to D298͑CF3O–H͒ϭ117.5Ϫϩ11..94 kcal/mol.
This is significantly closer to the high bond value originally
proposed by Wallington et al.,7 and backed by several ab
initio calculations,12,14,17,18 than to the low value of
ϳ109 kcal/mol suggested by Benson19 on the basis of group
additivity estimates. When error bars are included, then
C. ⌬Hؠf(CF3OF)
The fitted appearance potential of the CFϩ3 fragment
from CF3OF obtained in Sec. IV C is only an upper limit to
the true value. Using23,41 ⌬Hؠf 0͑OF͒ϭ25.9Ϯ2.3 kcal/mol
leads to ⌬Hؠf 0͑CF3OF͒ϾϪ184.6Ϯ3.4 kcal/mol. JANAF22
lists Ϫ 180.8 Ϯ 3.1, but their value was derived by reference
to ⌬Hؠf(CF2O), which has since been shown24 to be too low
by at least 4 kcal/mol. Also, there appears to be a problem37
with JANAF’s correction from 298 to 0 K.
D
298(CF3O–H)ϪD298͑HO–H͒ϭϪ1.8ϩϪ11..49 kcal/molϷ0 kcal/
mol. However, if one disregards the error bars, then
D(CF3O–H) is nominally slightly lower than D(HO–H).
This is in contrast to the original conclusion by Wallington
et al.,7 but perhaps in good accord with the more recent ob-
servation that CF3O does not appear to give rise to appre-
ciable hydrogen abstraction from water.8
A better approach to ⌬Hؠf(CF3OF) is to use experimen-
tal data evaluated by Batt and Walsh,20 which can be
combined42 to give the reaction enthalpy at 298 K for
CF3OF→CF2OϩF2 of 27.8 Ϯ 1.1 kcal/mol, somewhat lower
than the 30.1 Ϯ 3 kcal/mol adopted by JANAF.22 Our ap-
The appearance potential of CF2OHϩ from CF3OH leads
to
⌬Hؠf 298͑CF2OHϩ͒р84.2Ϯ1.0 kcal/mol,
which
is
J. Chem. Phys., Vol. 106, No. 22, 8 June 1997
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 193.0.65.67
On: Tue, 23 Dec 2014 13:45:23

9120
Asher et al.: Photoionization of trifluoromethyl compounds
13 M. Sana, G. Leroy, D. Peters, and C. Willante, J. Mol. Struct.
͑Theochem.͒ 164, 249 ͑1988͒.
ϳ28 kcal/mol higher than the value deduced from
PA͑CF2O͒, and may correspond to a different structure.
Photoionization of CF3OF produces a very weak parent,
with an apparent adiabatic IP͑CF3OF͒р12.710Ϯ0.007 eV,
significantly lower than previous values derived by photo-
electron spectroscopy.36 The threshold region of the CF3ϩ
fragment from CF3OF is suppressed by unfavorable Franck–
Condon factors for parent ionization, and a direct fit estab-
lishes only an upper limit of AP0͑CFϩ3 /CF3OF͒Ͻ13.38
Ϯ0.10 eV. However, the cusplike termination of growth in
the CF3OFϩ parent ion, at ϳ 12.97 Ϯ 0.01 eV, provides
a better estimate for the onset of the CFϩ3 fragment and
is compatible with the value of ⌬Hؠf 298͑CF3OF͒
ϭϪ176.9ϩϪ11..83 kcal/mol proposed here on the basis of other
arguments, which include an approximate determination of
AP0͑CF2Oϩ/CF3OF͒Ϸ14.2Ϯ0.1 eV. The latter appearance
potential is in excellent agreement with the reaction enthalpy
for CF3OF→CF2OϩF2 of 27.8 Ϯ 1.1 kcal/mol ͑298 K͒,
which can be derived from data evaluated by Batt and
Walsh.20
14 D. A. Dixon and R. Fernandez, Proceedings of the STEP-HALOCSIDE
AFEAS Workshop, Dublin, Ireland, March 1993 ͑unpublished͒.
15
͑a͒ C. W. Bock, M. Trachtman, H. Niki, and G. J. Mains, J. Phys. Chem.
98, 7976 ͑1994͒; ͑b͒ W. F. Schneider and T. J. Wallington, ibid. 99, 4353
͑1995͒; ͑c͒ C. W. Bock, M. Trachtman, H. Niki, and G. J. Mains, ibid. 99,
4354 ͑1995͒.
16
͑a͒ J. A. Montgomery, Jr., H. H. Michels, and J. S. Francisco, Chem. Phys.
Lett. 220, 391 ͑1994͒; ͑b͒ J. S. Francisco, ibid. 218, 401 ͑1994͒; ͑c͒ Chem.
Phys. 150, 19 ͑1991͒.
17 W. F. Schneider and T. J. Wallington, J. Phys. Chem. 98, 7448 ͑1994͒.
18 W. F. Schneider, T. J. Wallington, and M. D. Hurley, J. Phys. Chem. 98,
2217 ͑1994͒; W. F. Schneider, B. I. Nance, and T. J. Wallington, J. Am.
Chem. Soc. 117, 478 ͑1995͒.
19 S. W. Benson, J. Phys. Chem. 98, 2216 ͑1994͒.
20
͑a͒ L. Batt and R. Walsh, Int. J. Chem. Kinet. 14, 933 ͑1982͒; ͑b͒ 15, 605
͑1983͒.
21 W. F. Schneider, T. J. Wallington, and M. D. Hurley, J. Phys. Chem. 98,
2217 ͑1994͒.
22 M. W. Chase, C. A. Davies, J. R. Downey, Jr., D. J. Frurip, R. A. Mc-
Donald, and A. N. Syverud, JANAF Thermochemical Tables, 3rd ed. ͓J.
Phys. Chem. Ref. Data 14, Suppl. 1 ͑1985͔͒.
23 L. V. Gurvich, I. V. Veyts, and C. B. Alcock, Thermodynamic Properties
of Individual Substances ͑Hemisphere, New York, 1991͒, Vols. 1 and 2.
24 R. L. Asher, E. H. Appelman, and B. Ruscic, J. Chem. Phys. 105, 9781
͑1996͒.
The fitted value for the appearance potential of CF3ϩ
from CF3OCl, AP0͑CFϩ3 /CF3OCl͒ р 12.85 Ϯ 0.01 eV, leads
to ⌬Hؠf298͑CF3OCl͒уϪ175.6Ϯ1.0 kcal/mol, D298͑CF3–OCl͒
р88.4Ϯ0.3 kcal/mol, and D298(CF3O–Cl͒р52.8ϩϪ12..50 kcal/
mol. The latter value is in very good agreement with the
upper limit of Ͻ53 kcal/mol determined recently by Renga-
rajan et al.43
25 L. A. Curtiss, K. Raghavachari, C. P. Redfern, and J. A. Pople, J. Chem.
Phys. 106, 1063 ͑1997͒.
͑a͒ K. Seppelt, Angew. Chem. Int. Ed. Engl. 16, 322 ͑1977͒; ͑b͒ G. Kloter
26
¨
and K. Seppelt, J. Am. Chem. Soc. 101, 347 ͑1979͒.
͑a͒ D. E. Gould, L. R. Anderson, D. E. Young, and W. B. Fox, J. Am.
Chem. Soc. 91, 1310 ͑1969͒; ͑b͒ C. J. Schack and W. Maya, ibid. 91, 2902
͑1969͒; ͑c͒ M. Lustig, A. R. Pitochelli, and J. K. Ruff, ibid. 89, 2841
͑1967͒.
27
28 M. J. Frisch, G. W. Trucks, M. Head-Gordon, P. M. W. Gill, M. W.
Wong, J. B. Foresman, B. G. Johnson, H. B. Schlegel, M. A. Robb, E. S.
Replogle, R. Gomperts, J. L. Andres, K. Raghavachari, J. S. Binkley, C.
Gonzalez, R. L. Martin, D. J. Fox, D. J. Defrees, J. Baker, J. J. P. Stewart,
and J. A. Pople, GAUSSIAN 94, Rev. D. 1 ͑Gaussian Inc., Pittsburgh, PA,
1994͒.
ACKNOWLEDGMENT
This work was supported by the U.S. Department of En-
ergy, Office of Basic Energy Sciences, under Contract No.
W-31-109-ENG-38.
29 P. C. Hariharan and J. A. Pople, Theor. Chim. Acta 28, 213 ͑1968͒.
30
͑a͒ B. Ruscic and J. Berkowitz, J. Phys. Chem. 97, 11451 ͑1993͒; ͑b͒ J.
1
͑a͒ World Meteorological Organization Global Ozone Research and Moni-
Chem. Phys. 100, 4498 ͑1994͒; ͑c͒ 101, 7795 ͑1994͒; ͑d͒ 101, 7975
͑1994͒; ͑e͒ 101, 10936 ͑1994͒.
toring Project, Report No. 20, UN Environment Program, WMO, Geneva,
1990; ͑b͒ World Meteorological Organization Global Ozone Research and
Monitoring Project, Report No. 25, UN Environment Program, WMO,
Geneva, 1992.
31 R. L. Asher and B. Ruscic, J. Chem. Phys. 106, 210 ͑1997͒.
32 P. C. Haarhoff, Mol. Phys. 7, 101 ͑1963͒.
33 P. M. Wilt and E. A. Jones, J. Inorg. Nucl. Chem. 30, 2933 ͑1968͒.
34 Since Haarhoff’s expression ͑Ref. 32͒ does not readily accommodate a
hindered internal rotor, which would have been the most appropriate de-
scription for the internal energy associated with the CF3 torsion in
CF3OH, CF3OF, and CF3OCl, two versions of the internal energy distri-
bution were developed: one treated the internal rotation as a pseudovibra-
tion, and the other treated it as a free internal rotation. In all three cases
the difference between the two variations was only minor, and employ-
ment of either approach gave similar overall answers. The final fits were
performed using the pseudovibrator approach, which appeared to provide
a slightly better description of the internal energy distribution than the free
rotor model.
2
͑a͒ M. J. Molina and F. S. Rowland, Nature 249, 810 ͑1974͒; ͑b͒ R. S.
Stolarski and R. D. Rundel, Geophys. Res. Lett. 2, 443 ͑1975͒.
3 See, for example, E. W. Kaiser, T. J. Wallington, and M. D. Hurley, Int. J.
Chem. Kinet. 27, 205 ͑1995͒, and references therein.
4
͑a͒ M. K. W. Ko, et al., Geophys. Res. Lett. 21, 101 ͑1994͒; ͑b͒ A. R.
Ravishankara, A. A. Turnipseed, N. R. Jensen, S. B. Barone, M. Mills, C.
J. Howard, and S. Solomon, Science 71, 263 ͑1994͒.
5 See, for example: R. Meller and G. K. Moortgat, J. Photochem. Photobiol.
A: Chem. 86, 15 ͑1995͒, and references therein.
6 See, for example, T. J. Wallington and J. C. Ball, J. Phys. Chem. 99, 3201
͑1995͒, and references therein.
7 T. J. Wallington, M. D. Hurley, W. F. Schneider, J. Sehested, and O. J.
Nielsen, J. Phys. Chem. 97, 7606 ͑1993͒.
35 See, for example, B. Ruscic, J. Chem. Phys. 85, 3776 ͑1986͒; Modelling of
Structure and Properties of Molecules, edited by Z. B. Maksic ͑Horwood,
Chichester, 1987͒, pp. 221–238.
8 A. A. Turnipseed, S. B. Barone, N. R. Jensen, D. R. Hanson, C. J.
Howard, and A. R. Ravishankara, J. Phys. Chem. 99, 6000 ͑1995͒.
9 L. G. Huey, D. R. Hanson, and E. R. Lovejoy, J. Geophys. Res. 100,
18771 ͑1995͒; ͑b͒ W. F. Schneider, T. J. Wallington, and R. E. Huie, J.
Phys. Chem. 100, 6097 ͑1996͒.
36 S. G. Lias, J. E. Bartmess, J. F. Liebman, J. L. Holmes, R. D. Levin, and
W. G. Mallard, J. Phys. Chem. Ref. Data 17, Suppl. 1 ͑1988͒.
37 While preparing the function for this fit, an interesting discrepancy has
surfaced. JANAF ͑Ref. 22͒ states that it uses a hindered rotor approach
and spectroscopic data of Wilt and Jones ͑Ref. 33͒ to calculate the ther-
modynamic properties of CF3OF, and it lists 3.500 kcal/mol for H298
Ϫ H0 . Using the same data, we obtain 4.081 kcal/mol from the hindered
rotor approach. The more straightforward pseudovibrator approach
(56 cmϪ1) yields an almost identical 4.060 kcal/mol for the same quantity.
The internal energy function used in our fits was constrained to corre-
10
͑a͒ W. F. Schneider, T. J. Wallington, K. Minschwaner, and E. A. Stahl-
berg, Environ. Sci. Technol. 29, 247 ͑1995͒; ͑b͒ L. T. Molina and M. J.
Molina, Geophys. Res. Lett. 23, 563 ͑1996͒.
͑a͒ E. R. Lovejoy, L. G. Huey, and D. R. Hanson, J. Geophys. Res. 100,
11
18775 ͑1995͒; ͑b͒ L. G. Huey, P. W. Villalta, E. J. Dunlea, D. R. Hanson,
and C. J. Howard, J. Phys. Chem. 100, 190 ͑1996͒.
12 W. F. Schneider and T. J. Wallington, J. Phys. Chem. 97, 12783 ͑1993͒.
J. Chem. Phys., Vol. 106, No. 22, 8 June 1997
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 193.0.65.67
On: Tue, 23 Dec 2014 13:45:23

Asher et al.: Photoionization of trifluoromethyl compounds
9121
spond to the available internal energy that results from the hindered rotor
approach.
41 J. Berkowitz, P. M. Dehmer, and W. A. Chupka, J. Chem. Phys. 59, 925
͑1973͒.
38
42 Batt and Walsh ͑Ref. 20͒ list the following reaction enthalpies:
͑a͒ D. L. Osborn, D. J. Leahy, E. M. Ross, and D. M. Newmark, Chem.
Phys. Lett. 235, 484 ͑1995͒; ͑b͒ M. Meot-Ner ͑Mautner͒ and L. W. Sieck,
J. Phys. Chem. 90, 6687 ͑1986͒; ͑c͒ P. C. Engelking, G. B. Ellison, and
W. C. Lineberger, J. Chem. Phys. 69, 1826 ͑1978͒; ͑d͒ J. D. Garcia and J.
E. Mack, J. Opt. Soc. Am. 55, 654 ͑1965͒; ͑e͒ B. Ruscic and J. Berkowitz,
J. Chem. Phys. 95, 4033 ͑1991͒.
⌬Hؠr 298͑a͒ϭ24.5Ϯ0.7 kcal/mol
⌬Hr 298͑b͒ϭ46.8Ϯ0.5 kcal/mol
for
for
CF3OOCF3→CF3OFϩCF2O,
CF3OOCF3→2CF3O,and
⌬Hؠr 298͑c͒ϭ44.0 kcal/mol ͑presumably also
Ϯ
0.5 kcal/mol͒ for
CF3OF→CF3OϩF. The reaction enthalphy for CF3OF→CF2OϩF2 at 298
K is then 2⌬Hؠr 298(c) Ϫ ⌬Hؠr 298(b) ϩ ⌬Hrؠ 298(a) Ϫ D298(F2), or 27.8
39 S. G. Lias, J. F. Liebman, and R. D. Levin, J. Phys. Chem. Ref. Data 13,
695 ͑1984͒.
͑a͒ S. M. Collyer and T. B. McMahon, J. Phys. Chem. 87, 909 ͑1983͒; ͑b͒
Ϯ1.1 kcal/mol.
43 R. Rengarajan, D. W. Setser, and D. D. DesMarteau, J. Phys. Chem. 98,
10568 ͑1994͒.
40
C. E. Doiron and T. B. McMahon, Can. J. Chem. 59, 2689 ͑1981͒.
J. Chem. Phys., Vol. 106, No. 22, 8 June 1997
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 193.0.65.67
On: Tue, 23 Dec 2014 13:45:23