Products of the ultraviolet photodissociation of trifluoroacetic acid and acrylic acid

Article abstract of DOI:10.1039/a809064e

The photodissociation of trifluoroacetic and acrylic acids by the ultraviolet light from a flashlamp has been investigated by measuring the relative yields of some of the major products by time-resolved infrared absorption using tunable, narrow band diode lasers. Yields of CO₂ were measured both in the absence and presence of added O₂. The former experiments measure the CO₂ produced directly by decarboxylation of the acid, channel (2) below, the latter the sum of the yields from channels (1) and (2) since HOCO is rapidly converted to CO₂. The yields of CO from the decarbonylation channel (3) have also been measured. For trifluoroacetic acid, the relative yields are found to be [HOCO]: [CO₂]: [CO] = (0.28 ± 0.07): (0.61 ± 0.09): (0.11 ± 0.06) and, for acrylic acid, [HOCO]: [CO₂]: [CO] = (0.32 ± 0.08): (0.37 ± 0.08) :(0.31 ± 0.09). The results are discussed in relation to the other, limited, measurements on the photodissociation of these acids and whether these three processes are likely to occur independently of one another.

Full text of DOI:10.1039/a809064e

View Article Online / Journal Homepage / Table of Contents for this issue
Products of the ultraviolet photodissociation of triÑuoroacetic acid
and acrylic acid
Michael C. Osborne, Qiang Li and Ian W. M. Smith*
School of Chemistry, T he University of Birmingham, Edgbaston, Birmingham, UK B15 2T T
Received 19th November 1998, Accepted 28th January 1999
The photodissociation of triÑuoroacetic and acrylic acids by the ultraviolet light from a Ñashlamp has been
investigated by measuring the relative yields of some of the major products by time-resolved infrared
absorption using tunable, narrow band diode lasers. Yields of CO were measured both in the absence and
2
presence of added O . The former experiments measure the CO produced directly by decarboxylation of the
2
2
acid, channel (2) below, the latter the sum of the yields from channels (1) and (2) since HOCO is rapidly
converted to CO . The yields of CO from the decarbonylation channel (3) have also been measured.
2
R (\CF or C H )COOH ] hl ] R ] HOCO
(P.1)
(P.2a, P.2b)
(P.3a, P.3b)
3
2 3
]
]
R ] H (or RH) ] CO
2
R ] OH (or ROH) ] CO
For triÑuoroacetic acid, the relative yields are found to be [HOCO] : [CO ] : [CO] \ (0.28 ^ 0.07) : (0.61
2
^
^
0.09) : (0.11 ^ 0.06) and, for acrylic acid, [HOCO] : [CO ] : [CO] \ (0.32 ^ 0.08) : (0.37 ^ 0.08) : (0.31
2
0.09). The results are discussed in relation to the other, limited, measurements on the photodissociation of
these acids and whether these three processes are likely to occur independently of one another.
In recent years there have been a number of spectroscopic1,2
and kinetic3,4 studies of HOCO radicals in which they were
generated by photodissociation of a carboxylic acid, usually
The thermodynamics of these and other photodissociation
channels in triÑuoroacetic and acrylic acids are given in Table
1.
acrylic acid (CH xCHCOOH). The HOCO radical is a tran-
The Ðrst study of the photolysis of triÑuoroacetic was
carried out by Mearns and Back in 1963.12 They used a con-
tinuous medium pressure mercury arc to photolyse the acid
between 200 and 220 nm. The acid vapour pressure was
varied between 10 and 100 Torr and the temperature between
90 and 190 ¡C. The major products were CO and C F with
2
sient intermediate in the important reaction between OH rad-
icals and CO,5h8 and in the reverse reaction,
OH ] CO ½ H ] CO
(R.1)
2
and can indeed be stabilised in the case of the forward reac-
tion, if the total pressure in the system is sufficiently high.9,10
Amongst other results, measurements on the reactions of
2
2 6
smaller yields of CO, H and CF H. To examine the mecha-
2
3
nism further, Mearns and Back analysed the products of the
HOCO have shown it to react rapidly with O .3,4,11
photoinitiated reaction in the presence of isobutane, C F
2
4 10
and but-1-ene. C F had only a minor e†ect on the product
4 10
HOCO ] O ] HO ] CO
(R.2)
yields, suggesting that the relaxation of vibrationally excited
2
2
2
species was not the cause of the marked changes which
occurred when isobutane or but-1-ene was added. The pres-
ence of C F and its replacement by CF H in the presence of
Determinations of the rate constant for this reaction yield
values of k varying from 1.45È1.9 ] 10~12 cm3 molecule~1
2
s~1 at 298 K.3,4,11 One conclusion of these measurements is
2
6
3
isobutane appeared to establish the production of CF rad-
that the nett result of reaction between OH and CO in the
3
icals in the primary photochemical process beyond doubt,
atmosphere is to yield HO ] CO , either via reaction (R.1)
2
2
although all three processes (P.1), (P.2a) and (P.3a), with R \
plus the combination of the H atom product with O or via
2
CF , yield CF .
the stabilisation of the HOCO formed from OH ] CO fol-
3
3
Largely on the basis of the e†ects of added isobutane and
lowed by reaction (R.2).
but-1-ene on the yields of CO and CO, Mearns and Back
In contrast to this rather well-understood situation, many
of the important details of the ultraviolet photodissociation of
the carboxylic acids are not well-known and there have been
relatively few experimental studies of these processes using
modern techniques. Among the photodissociation channels
that have been postulated are
2
postulated that the CwC bond Ðssion represented by eqn.
(P.1) was the principal primary photochemical process. Of
course, the evidence for the production of HOCO radicals was
only indirect and the characterisation of (P.1) as the major
photodissociation channel for triÑuoroacetic acid on the basis
of these experiments is open to some doubt. However, before
the present investigation, there had been no direct study of the
photodissociation of triÑuoroacetic acid reported in the liter-
ature.
As mentioned earlier, the photodissociation of acrylic acid
has been the usual source for spectroscopic and kinetic studies
of the HOCO radical. In addition, there have been a few
recent studies of the photodissociation itself, though none has
RCOOH ] hl ] HOCO ] R
(P.1)
(P.2a)
(P.2b)
(P.3a)
(P.3b)
]
]
]
]
CO ] R ] H
2
CO ] RH
2
CO ] R ] OH
CO ] ROH
Phys. Chem. Chem. Phys., 1999, 1, 1447È1454
1447

View Article Online
Table 1 Thermodynamics of di†erent channels in the photodissociation of triÑuoroacetic and acrylic acids
R \ CF
R \ CH xCH
3
2
Products
* H/kJ mol~1
M* H [ (N hl)N/kJ mol~1
* H/kJ mol~1
M* H [ (N hl)N/kJ mol~1
r
r
A
r
r
A
R ] HOCO
R ] H ] CO
]339
]385
[60
[231
[185
[630
[81
]414
]461
[4
[156
[109
[574
[5
2
RH ] CO
2
R ] OH ] CO
ROH ] CO
RCO ] OH
RCHO ] O
]489
]565
]89
[481
[134
[73
]435
]496
For each acid, the Ðrst column shows the standard enthalpy change for transformation of RCOOH to the stated products at 298 K. The second
column shows the enthalpy change associated with photodissociation by photons of wavelength of 210 nm (N hl \ 596.6 kJ mol~1).
A
had the determination of the branching ratios into di†erent
product channels as their primary aim.
and relaxation of vibrationally excited HOCO radicals. The
rates involved mean that the dissociation would have to be
Rosenfeld and Weiner13 looked for infrared Ñuorescence
following the pulsed excimer laser irradiation of acrylic acid at
due to tunnelling through the barrier leading to H ] CO .
2
This suggestion is further discussed later in this paper.
1
93, 248 and 308 nm. No emission was detected at the longest
In the experiments described in the present paper, we have
used a time-resolved infrared absorption technique to deter-
mine the branching ratios into channels (P.1), (P.2), and (P.3)
wavelength. At the two shorter wavelengths, emission at 4.3
lm, corresponding to the l band region of CO , was
observed and experiments with cold gas Ðlters showed that a
substantial fraction of the emission came from levels with
3
2
when triÑuoroacetic acid (R \ CF ) and acrylic acid (R \
3
CH xCH) are photolysed by the output from a conventional
2
v [ 1. Rosenfeld and Weiner could detect no emission in the
Ñashlamp. To measure the relative yields of HOCO and CO ,
3
2
CwH stretching region (ca. 3.3 lm) or from CO at 4.8 lm.
absorbances on lines in the (0001,0000) band of CO were
2
measured in the presence and absence of O , HOCO being
2
However, in assessing these results, it is necessary to remem-
ber that the Einstein coefficients for these di†erent systems
converted rapidly to CO by reaction (R.2) when O was
2
2
di†er appreciably, with that for the (0001È0000) band of CO
ca. 14 times that for the (1, 0) band of CO. Rosenfeld and
Weiner concluded that decarboxylation, i.e. channel (P.2), is a
major process and forms CO excited in the l vibration.
Lessard14 followed up the work of Rosenfeld and Weiner
using time-resolved infrared absorption in a manner similar to
that reported in the present paper. Searches were made for
both CO and CO
CH xCHCOOH at 193 and 248 nm. CO was only observed
using the shorter wavelength, though in assessing this result
and those of Rosenfeld and Weiner, one should note that
acrylic acid has only a very small absorption cross-section at
the longer wavelength15 and therefore any concentration of
vibrationally excited CO might have been below their detec-
tion limit. Prompt formation of CO was observed at both
wavelengths, suggesting its production in a primary process.
The most dynamical study to date of the photodissociation
of acrylic acid is that performed by Kitchen et al.16 in a
crossed laser-molecular beam apparatus. Products were
detected by mass spectrometry using 200 eV electron impact
ionisation. The measured mass peak at m/e \ 44 was attrib-
uted entirely to the CO ` daughter ion of HOCO. Kitchen et
al. concluded that, at 193 nm, neither decarboxylation,
channel (P.2), nor decarbonylation, channel (P.3), is important
added. The background source for these measurements was a
2
diode laser tuned to the centre of an appropriate line in the l
3
fundamental band of CO . Separate measurements, using a
2
second diode, were carried out on the CO produced in
2
3
channel (P.3). In all cases, the absorbances were measured at
delays sufficient to ensure that both reaction (R.2) and the
relaxation of any vibrationally excited CO
complete.
2 and CO were
following laser photolysis of
2
2
Experimental method
Fig. 1 provides a schematic of the apparatus used in the
present experiments. Weak mixtures of the carboxylic acid in
helium or argon were passed through a tubular reactor of 18
mm internal diameter. The central portion of this tube is made
of Spectrosil quartz and a length of ca. 80 cm is surrounded
by an annular Ñash lamp. The tube is Ðtted at both ends with
2
CaF windows which are set at 35¡ to minimise any e†ects
2
due to back reÑection of the radiation from the diode laser.
The annular space between the outer surface of the Spectro-
sil reactor and the outer quartz sleeve of the Ñashlamp is ca. 8
mm thick and is Ðlled with ca. 10 Torr of xenon. Using “OÏ-
rings, this volume is sealed by brass end-pieces which also
serve as the electrodes for the Ñashlamp. The Ñashlamp forms
part of a custom-built charging circuit which includes two 0.5
lF, 20 kV capacitors and a power supply capable of charging
at a rate of 1 kJ s~1. The capacitor is generally charged to
between 13 and 17 kV yielding a dissipated energy of between
85 and 145 J when the lamp is Ðred. The lamp was Ðred at
rates between 0.6 and 1.0 Hz. To ensure that there was no
build up of products, the gas Ñow was adjusted so that the
reaction cell was re-Ðlled twice during the period between
Ñashes.
In the present experiments, the main disadvantage of using
a Ñashlamp rather than a pulsed laser to e†ect photo-
dissociation was the lack of deÐnition of the energy of the
photons absorbed by the molecules of carboxylic acid. Thus,
photolysis occurs over a range of wavelengths with a distribu-
tion of input energies which is the result of the overlap of
lamp output, limited at short wavelengths (ca. 165 nm) by the
transmission of Spectrosil quartz, with the absorption spec-
2
and that only CwC bond scission, channel (P.1), and CwO
bond Ðssion leading to CH xCHCO ] OH are signiÐcant
2
photodissociation channels.
Finally, in this survey of previous work, we mention the
work of Miyoshi et al.4 Although their primary aim was to
make kinetic measurements on reactions of HOCO using a
time-resolved mass spectrometric technique, they used photo-
dissociation of acrylic acid to generate HOCO and they
observed how the concentration of HOCO changed with time
both in the presence and absence of added co-reagent.
Because they used photoionisation with hl 4 10.03 eV, they
could observe only a limited range of species, and peaks at
m/e \ 44 and m/e \ 55 were assigned to C H OH and
2
3
C H CO, respectively. It was further suggested that the decay
2
3
of HOCO in the absence of any added reagent at a rate which
was both non-exponential and dependent on total pressure
represented competition between unimolecular dissociation
1448
Phys. Chem. Chem. Phys., 1999, 1, 1447È1454

View Article Online
Fig. 1 Apparatus used to study the products of the ultraviolet photodissociation of triÑuoroacetic and acrylic acids using time-resolved infrared
absorption spectroscopy.
trum of the acid. We shall return to this point in the Dis-
cussion section of the paper. As, in the present work, the aim
was to compare the yields of Ðnal products, the timescale of
the measurements was quite long (see below) and consequent-
ly the longer duration of the light output from the lamp
detector and the output from the ampliÐer was passed to a
digital storage oscilloscope. Generally ca. 20 traces were aver-
aged before subsequent processing. The time response of the
detectorÈampliÐer combination was about 10 ls in the experi-
ments that are reported in the present paper.
(
measured as about 10 ls) compared with a laser was unim-
The output frequency of the diode laser could be estimated
portant. One advantage of a Ñashlamp is that it provides
rather uniform illumination of the reaction volume, thereby
minimising e†ects resulting from di†usion of the species being
observed.
within a few wavenumbers using the monochromator.
However, to Ðnd the frequency to the accuracy required it was
necessary to use low pressures of one or more calibration
gases, either in the reaction vessel or in a separate cell,
together with a high Ðnesse germanium transmission etalon
(Mutek GmbH, MDS 1920). The calibrant gases used were
CO , CO, OCS and N O (including their isotopomers). The
The tunable, infrared “probeÏ radiation is provided by a
liquid N cooled diode laser system (Mutek GmbH, MDS
2
1
150). Two di†erent diodes (one for measurements on CO ,
2
2
2
one for measurements on CO) were employed in the experi-
ments that are reported here. Because of the divergence of the
radiation from a diode laser, collimating optics must be
employed to generate a suitable probe beam. The collimating
optics that were used originally (Mutek GmbH, MDS 1125)
consisted of a total of Ðve gold-coated mirrors, 3 planar, 1
ellipsoidal and 1 toroidal, mounted on a moveable stage. This
arrangement produced a collimated beam of ca. 14 mm diam-
eter and a divergence of about 1.5 mrad. It was later replaced
by a single 90¡ o†-axis parabolic mirror which gave a beam of
similar diameter and divergence but much improved intensity,
leading to better signal-to-noise. After the probe beam
emerges from the reaction vessel, it is focused onto the
entrance slit of a monochromator (HilgerÈWatts, D330) which
is Ðtted with a di†raction grating blazed at 6 lm. This mono-
chromator is able to select a single mode from the diode laser
infrared absorption frequencies of these gases have all been
assigned, tabulated and provided in diagrammatic form.17,18
Of course, in the present work, tuning of the diode lasers for
time-resolved absorption measurements was made easier by
the fact that we were observing stable species (CO and CO)
2
in levels with high thermal populations.
The reaction cell and Ñashlamp are attached to a standard
gas handling line constructed from Pyrex and Ðtted with
greaseless stopcocks. All Ñows of gases were set using mass
Ñow controllers (MKS and Bronkhurst Hi-Tec). The helium
or argon carrier gas was taken directly from a cylinder, He
(Air Products, GC grade) and Ar (Air Products, Resolution
grade), and was used without further puriÐcation. Oxygen was
also delivered directly from a cylinder (BOC, 99.6%).
Di†erent methods of delivery had to be used for the two
carboxylic acids. The vapour pressure of triÑuoroacetic acid is
sufficiently high that dilute mixtures could be prepared with
the carrier gas in a 20 l bulb and a controlled Ñow could be
delivered using an appropriate mass Ñow controller. In con-
trast, because of its low vapour pressure at room temperature
(only 3.5 Torr), acrylic acid was delivered using a bubbler fol-
lowed by a cold trap. By controlling the temperature of the
which is directed onto a liquid N cooled InSb photovoltaic
2
detector (Infrared Associates) placed at the exit slit of the
monochromator. Because of the presence of CO in the atmo-
2
sphere it is essential to enclose and purge the complete path of
the probe laser. This is done using N from a cylinder. An AC
2
coupled ampliÐer was used in conjunction with a cooled InSb
Phys. Chem. Chem. Phys., 1999, 1, 1447È1454
1449

View Article Online
cold trap, and hence the partial pressure of the acid at this
point, and the Ñow of carrier gas through this line, it was pos-
sible to control the partial pressure of the acrylic acid in the
reaction cell. Each new sample of acid was degassed by bub-
bling helium through it.
The general form of all the traces that have been recorded is
similar. Following a prompt rise, which on close examination
seems of larger amplitude in triÑuoroacetic acid than in
acrylic acid, there is a more gradual rise to an eventual steady
state value which we take to be a measure of the total yield of
In calculating the amount of carboxylic acid in the reaction
cell and the fraction of the acid present as monomer, it was
necessary to allow for dimerisation. For triÑuoroacetic acid,
the equilibrium constant for dimer ½ 2 monomer is known to
CO under the conditions of the experiment. The slow rise
from the prompt signal to the absorbance at long time is
2
taken to reÑect the Ðnite time for relaxation of CO molecules
2
formed in vibrationally excited states. In addition, in the pres-
be K \ 4.14 Torr.19 Thus at 300 mTorr, the highest pressure
ence of O , there is a delay in the conversion of HOCO
p
2
used in the present experiments, 6.3% of the species are
formed in process (P.1) to CO via reaction (R.2). However, in
2
dimers. Allowance also had to be made for the higher fraction
of dimers present in the triÑuoroacetic acid originally mea-
sured into the storage bulb from which acid was delivered.
The corresponding equilibrium constant for acrylic acid is
the presence of 0.5 Torr of O , the Ðrst-order rate constant
2
k
[O ] is equal to ca. 3 ] 104 s~1 indicating rapid reaction
compared with the time taken for the absorbance to reach its
long time value. The period of several milliseconds over which
the absorbances remain constant at long delay times conÐrms
R.2
2
not known. We estimate a value of K \ 2.6 Torr based on a
p
comparison of K with pK values for a sequence of carbox-
that the stable product, CO , is formed rather uniformly
p
a
2
ylic acids. Here, experiments are reported for pressures of
acrylic acid up to 95 mTorr where 3.5% of the molecules are
dimers. Again allowance was made for the higher fraction of
dimers present in the Ñow line through which acrylic acid was
admitted.
through the reaction cell, and that no CO is formed by unac-
2
counted for secondary chemistry.
ConÐrmation that vibrational relaxation is responsible for
the delay in the measured absorbances reaching their Ðnal
values is obtained if one considers the experimental conditions
and the rate constants for relaxation by the various species
present in the experimental mixtures. In most of the present
experiments, helium was present to bring the total pressure to
Further details of the equipment and methods used in the
present experiments are available elsewhere.20
1
0 Torr. (For reasons explained later, some measurements
Results
were carried out on mixtures diluted to 50 Torr in He.)
Traces of transmitted intensity vs. time were recorded follow-
ing the photolysis of triÑuoroacetic acid and acrylic acid with
the diode laser tuned to the central frequency of the P(2),
Helium efficiently relaxes the l /l vibrational manifold of
1
2
CO . Thus, for CO (0110), the rate constant for relaxation is
2
2
equal to 1.1 ] 10~13 cm3 molecule~1 s~1,21 corresponding to
a Ðrst-order rate constant of 3.7 ] 104 s~1 at 10 Torr of He.
This suggests that the gradual increase in absorbance is only
due to CO molecules excited in the l manifold which is not
2
0
347.576 cm~1, and P(4), 2345.985 cm~1, lines in the (0001,
000) band of CO . These signals were converted to absorb-
2
ances and a few examples of the many traces that have been
recorded are displayed in Fig. 2. In each of these diagrams,
2
3
strongly coupled to the l /l levels.
1
2
traces are compared for experiments in the absence of O and
Both spontaneous emission and collisions with helium can
2
in the presence of 0.5 Torr of O .
make a signiÐcant contribution to the l relaxation. The Ein-
2
3
Fig. 2 Typical traces of absorbance vs. time for CO formed in the presence (upper trace) and absence (lower trace) of 0.5 Torr of O . All four
2
2
measurements were made on the P(2) line of the (0001,0000) band of CO , (a) 310 mTorr of triÑuoroacetic acid; (b) 136 mTorr of triÑuoroacetic
2
acid; (c) 53 mTorr of acrylic acid; and (d) 25 mTorr of acrylic acid. All these experiments were carried out at a total pressure of 10 Torr in helium
diluent. An absorbance of 0.1 on the P(2) line of the (0001,0000) band of CO corresponds to a total CO concentration of 9.3 ] 1012 molecule
2
2
cm~3.
450
1
Phys. Chem. Chem. Phys., 1999, 1, 1447È1454

View Article Online
stein coefficient for the former process is 420 s~1.22 The
second-order rate constant of relaxation by He is21,23
Ðgure is (53 ^ 8)% in the case of acrylic acid. In both cases,
the branching ratio was independent of the acid concentration
and the total pressure over the ranges studied.
In order to examine the branching ratio into process (P.3),
similar experiments to those just described for CO2 were
carried out on lines in the (1, 0) band, mainly on the R(8) line
at 2176.284 cm~1. Examples of the variation of absorbance
with time following the photodissociation of triÑuoroacetic
acid and of acrylic acid are shown in Fig. 4. They show a
2
.4 ] 10~15 cm3 molecule~1 s~1 yielding a pseudo-Ðrst-order
rate coefficient of 770 s~1 at a helium pressure of 10 Torr. In
the absence of any other relaxation channels, CO (0001)
2
would have a relaxation time of ca. 0.84 ms. Measurements
which will be described in a future publication20,24 yield rate
constants for relaxation of CO (0001) by triÑuoroacetic acid
2
and acrylic acid of (5.6 ^ 1.9) ] 10~12 and (6.9 ^ 2.3) ] 10~12
cm3 molecule~1 s~1, respectively. The absolute value of these
rate constants, and their similarity, suggest that relaxation
may occur by vibration-to-vibration (VÈV) transfer to the CO
vibration in the acid.
similar form to the CO traces: a prompt signal, followed by a
2
slow rise to a Ðnal steady value. The poorer signal-to-noise,
especially in the case of the trace from triÑuoroacetic acid, is
the result of three factors: the small yield of CO, the small
The relative importance of the channels (P.1) and (P.2)
absorption strength relative to CO , and the noise from the
Ðring of the Ñashlamp a†ecting the performance of the diode.
2
yielding HOCO and CO can be determined by comparing
2
absorbances on a given CO (0001,0000) absorption line at
Again, the slow rise is attributed to the relaxation of CO
from vibrational states above v \ 0. Comparison of the
absorbance at long time with that measured promptly sug-
gests that when triÑuoroacetic acid is photolysed ca. 40% of
the CO is produced in v \ 0, the remainder in levels with
v [ 0. In the case of acrylic acid, these Ðgures are approx-
imately reversed with ca. 60% in v \ 0, 40% in v [ 0.
Vibrationally excited CO is notoriously impervious to col-
lisional relaxation, at least by atomic collision partners. The
Ðrst-order rate constants obtained by adding the Einstein
2
long delay times. Inspection of the traces in Fig. 2 shows: (a)
that photolysis of acrylic acid gives higher absolute yields
than that of triÑuoroacetic acid (note the higher absorbances
from smaller acid concentrations in the former case), and (b)
that, in the presence of O , a higher proportion of CO comes
2
2
directly from photolysis in the case of triÑuoroacetic acid. The
higher absolute yields from acrylic acid are consistent with its
stronger absorption at the ultraviolet wavelengths above ca.
1
65 nm25 which are produced by the Ñashlamp and transmit-
ted by the Spectrosil reaction vessel.
We have examined the yields of CO as a function of both
2
the partial pressure of acid and the pressure of added helium.
The results are shown in Fig. 3. From these data, we calculate
that, in the presence of O , (70 ^ 5)% of the CO from tri-
2
2
Ñuoroacetic acid comes directly from photolysis, whereas this
Fig. 4 Typical traces of absorbance vs. time for CO: (a) from 180
mTorr of triÑuoroacetic acid measured on the (1, 0) P(5) line; (b) from
53 mTorr of acrylic acid measured on the (1, 0) R(8) line, and (c) from
25 mTorr of acrylic acid measured on the (1, 0) R(8) line. All these
experiments were carried out at a total pressure of 10 Torr in helium
diluent. An absorbance of 0.1 on the P(5) line of the (1, 0) band of CO
corresponds to a total CO concentration of 1.4 ] 1013 molecule
cm~3; and absorbance of 0.1 on the R(8) line to a total CO concentra-
tion of 1.6 ] 1013 molecule cm~3.
Fig. 3 Proportion of the CO produced in the presence of added O
2
2
which can be attributed to direct formation in channel (P.2): (a) from
triÑuoroacetic acid, and (b) from acrylic acid. The closed symbols rep-
resent measurements made in 10 Torr of helium, the open symbols in
5
0 Torr of helium. The triangles are data derived from measurements
on the (0001,0000) P(2) line, the squares and circles from the (0001,
000) P(4) line.
0
Phys. Chem. Chem. Phys., 1999, 1, 1447È1454
1451

View Article Online
rather than to the less exothermic radical products, means
that our estimate of the temperature rise is an upper limit and
conÐrms that thermal e†ects can be ignored in analysing our
experiments.
Finally, we considered the possibility that all or some frac-
tion of the observed CO and CO products was formed by
2
photodissociation of HOCO radicals created in the primary
photolysis of the carboxylic acid. Unfortunately, there is no
published ultraviolet spectrum of HOCO to aid in the assess-
ment of this possibility. To test it experimentally, yields of
CO were measured in the presence and absence of added O
2
2
with di†erent energies dissipated through the Ñashlamp. If
photolysis of HOCO was producing signiÐcant concentrations
of CO , this e†ect would be greatest at the largest Ñashlamp
2
energies and the measured branching ratios would change as
the Ñashlamp energy was varied. Fig. 5 shows results of these
experiments on both triÑuoroacetic and acrylic acid. In both
cases, the branching ratio is invariant with the Ñashlamp
energy and we conclude that photolysis of HOCO is not a
signiÐcant source of CO (and, by implication, of CO) under
2
the conditions of our experiments.
Discussion
The present work seems to establish unequivocally that, for
the photolysis wavelengths used in our experiments, photo-
dissociation of triÑuoroacetic acid and acrylic acid both
proceed by at least three competing channels: (P.1) cleavage of
the CÈC bond to yield R ] HOCO; (P.2) decarboxylation, i.e.
Fig. 5 Proportion of the CO produced in the presence of added O
2
2
loss of CO with RH or R ] H as co-products; and (P.3)
which can be attributed to direct formation in channel (P.2) in experi-
ments performed with di†erent energies dissipated through the Ñash-
lamp: (a) from triÑuoroacetic acid, and (b) from acrylic acid. All
measurements were made on the (0001,0000) P(4) line. The Ñash ener-
gies were (|) 115 J, ()) 33 J and (L) 21 J. In all cases the total
pressure was 10 Torr.
2
decarbonylation, i.e. loss of CO with R ] OH or ROH as co-
products. The relative branching ratios for these three chan-
nels are given in Table 2. In comparing our quantitative
Ðndings with those of others, it is, of course, necessary to bear
in mind the di†erent experimental conditions, especially the
photolysis wavelength(s). In other modern investiga-
tions,4,13,14,16 photodissociation has been initiated with a
pulsed laser, usually at 193 nm, rather than a Ñashlamp which
produces ultraviolet light over a range of wavelengths down
to a limit determined by the transmission of Spectrosil quartz.
Within the wavelength range emitted by the Ñashlamp and
transmitted by Spectrosil quartz, triÑuoroacetic acid exhibits
coefficient for spontaneous emission26 to the pseudo-Ðrst-
order rate constant for relaxation by He27 is only 34 s~1.
There seems little doubt that, in our experiments, relaxation of
CO (v [ 0) is dominated by collisions with the acid. This is
consistent with the more rapid rise of absorbance with time
evident in Fig. 4(b), 53 mTorr of acrylic acid, compared with
4
(c), where the acrylic acid pressure is 25 mTorr. The form of
two absorption bands: a weak system (j
peak
D 45 dm3 mol~1 cm~1) and a stronger system to shorter
D 205 nm;
the traces in Fig. 4 suggests that the rate constants for relax-
ation of CO (v [ 0) by the two acids lie in the range 10~12 to
1
e
peak
wavelengths (j
D 160 nm; e
D 2 ] 103 dm3 mol~1
peak
peak
0~13 cm3 molecule~1 s~1. Relaxation is again likely to occur
cm~1).25 The former has been assigned25 to a nÈp* transition
on the carbonyl group; the latter to a nÈ3s Rydberg tran-
sition. The nÈp* band shows a little structure but the band at
shorter wavelength is completely unstructured.
by VÈV transfer to the CO stretching mode in the acid.
Assuming the absorption lines to have a Doppler proÐle,
the measured absorbances can be converted to concentrations
of CO2 and CO, using the well-established integrated inten-
sities for the (0001,0000) and (1, 0) bands in these molecules.28
The resultant factors, which have been used to convert
absorbances to concentrations, are given in the captions to
Figs. 2 and 4. We found that absolute yields of CO and CO
do not scale linearly with the acid concentration in the case of
either acid, which we attribute to the fact that absorption
occurs over a range of ultaviolet wavelengths with di†erent
absorption cross-sections. The total observed product concen-
tration, i.e. [HOCO] ] [CO ] ] [CO], expressed as a frac-
tion of the initial concentration of the acid is found from the
measured yields of these products to be \5% in all cases and
much less in most experiments. This Ðgure can then be used to
estimate the amount of energy released in the photochemical
In acrylic acid, the weak nÈp* transition is shifted to longer
wavelengths, starting at ca. 280 nm and merging into a strong-
er system below ca. 220 nm.15 This stronger band (j
D 185
peak
D 5.5 ] 103 dm3 mol~1 cm~1) was originally assign-
nm; e
peak
ed as a pÈp* transition within the vinyl group.15 However,
based on a study of emission from acrylic acid excited at 199
nm, Arendt et al.29 concluded that absorption of a photon in
that region promotes an electron to a p* orbital which is a
mixture of p* (CxC) and p* (CxO) and that this state then
predissociates. Others have shown that excitation of acrylic
acid within this same band, but at 193 nm, gives rise to
HOCO (among other photochemical products).2,4
2
2
The ultraviolet light from the Ñashlamp used in our experi-
ments would excite each acid in the two bands described in
the previous paragraphs. The absolute yields of photo-
chemical products were found not to vary linearly with the
partial pressure of the acid, suggesting that at acid concentra-
tions lower than those present in the current experiments,
absorption in the lower wavelength system was probably pre-
dominant. At higher acid concentrations, signiÐcant excitation
probably occurs in both absorption bands. We note that the
processes (P.1), (P.2) and (P.3), and in the HOCO ] O reac-
2
tion when O is present, and hence the consequent tem-
2
perature rise. Assuming that photodissociation occurs only via
channels (P.1), (P.2b) and (P.3b), based on the exothermicities
given in the last column of Table 1, we estimate that any tem-
perature increase is always less than 5 K. The assumption that
channels (P.2) and (P.3) proceed to the molecular products,
1452
Phys. Chem. Chem. Phys., 1999, 1, 1447È1454

View Article Online
relative yields of HOCO, CO and CO do not, within experi-
mental error, depend on the concentrations of triÑuoroacetic
(decarboxylation) nor loss of CO (decarbonylation) is a signiÐ-
cant primary photodissociation channel.
2
or acrylic acid through the ranges of partial pressure
employed in these measurements.
As mentioned in the Introduction, the only paper on tri-
Ñuoroacetic acid which seems relevant to the present study is
that of Mearns and Back.12 They photolysed relatively high
partial pressures of the acid at temperatures between 90 and
It is possible that the di†erence between the conclusion of
Kitchen et al.16 that only primary CwC and CwO bond Ðs-
sions are major pathways, and our own conclusion that
photodissociation occurs at comparable rates via pathways
(P.1), (P.2) and (P.3), is entirely due to di†erences in the experi-
mental conditions, especially the photolysis wavelength.
However, this seems unlikely and we have examined possible
alternatives. First we note that the exothermicities of channels
(P.1) and (P.3a) di†er by 151 kJ mol~1,30 almost exactly the
energy of the excited state of HOCO invoked by Kitchen et
al.16 to explain the slower moving peaks in the time-of-Ñight
mass spectra for m/e \ 17 (and m/e \ 44). Of course, the
translational energy distribution of any OH formed in a
process, like (P.3a), generating three fragments will depend on
the details of the energy release but the energy di†erence
between these pathways is, at least, highly suggestive.
190 ¡C using light between 200 and 220 nm. Perhaps their
clearest and most valuable result was their observation among
the products of C F , providing strong evidence for the for-
2
6
mation of CF radicals via steps (P.1), (P.2a) and (P.3a). No
3
CF OH was reported and although CF H was detected it
could have been formed indirectly, for example by reaction of
CF radicals with CF COOH.
3
3
3
3
On the basis that addition of isobutane reduced the yields
of CO and CO, Mearns and Back concluded that (P.1) was
2
the main primary process in the photolysis of triÑuoacetic
acid. However, we note that the yields of CO and CO were
only reduced by addition of isobutane, to apparently steady
Of course, channel (P.3) does not produce CO and rela-
2
2
tively slow moving species were also observed at m/e \ 44 in
the experiments of Kitchen et al.16 Our experiments indicate
appreciable decarboxylation via channel (P.2) and it seems
yields, and that the CO : CO ratio at high concentrations of
2
isobutane was the same as that which we determine directly,
within experimental error. It seems likely that, in Mearns and
BackÏs experiments in the absence of isobutane, radicals
formed in the primary process could undergo reactions yield-
ing certainly CO , e.g. CF ] CF COOH ] CF ] CO
that the CO produced in this channel might be responsible
for the slower-moving species at m/e \ 44, despite the fact that
2
the exothermicity of (P.2a) is not a lot less than that of (P.1), if
the light H atom were to carry o† a major portion of the
energy released in this process. For example, if process (P.2a)
occurs by successive CwC and OwH bond breakages, much
2
3
3
3
2
]
CF H, and possibly CO. We agree with Mearns and Back
3
that (P.1) is a major channel in the ultraviolet photolysis of
CF COOH but not that it is the major channel.
of the energy released as HOCO fragments to CO ] H must
3
2
Our results for acrylic acid are broadly in agreement with
inevitably go into the motion of the H atom.31 Therefore we
those of Rosenfeld and Weiner13 and of Lessard.14 Our data
show strong evidence for the production of the vibrationally
excited CO2 which was observed by Rosenfeld and Weiner
from its infrared Ñuorescence. They did not observe emission
from CO, but it should be noted that the emission coefficient
for the (1, 0) band of CO is ca. 14 times smaller than that for
speculate whether the slower moving species observed at
m/e \ 44 in the experiments of Kitchen et al. could be CO
2
molecules created in channel (P.2a). What is certain is that
further investigation of the photodissociation of carboxylic
acids is required to pin down the primary processes occurring
and whether these change with the wavelength of the photoly-
sis source.
Miyoshi et al.4 compared the kinetic behaviour of HOCO
radicals formed from photodissociation of acrylic acid at 193
nm with that of HOCO formed as a product of the almost
thermoneutral reaction between Cl atoms and formic acid. In
the presence of any added reagent, loss was more rapid in the
former case. Moreover, the loss did not obey Ðrst-order
kinetics and it was faster at lower total pressures. Miyoshi et
al.4 attributed this accelerated loss rate to dissociation of
vibrationally excited HOCO radicals by tunnelling through
(
0001, 0000) band of CO so emission from vibrationally
2
excited CO may have been below their detection limits.
Lessard14 observed both CO and CO by kinetic infrared
2
absorption spectroscopy when acrylic acid was photolysed at
193 nm. He measured signiÐcant production of vibrationally
excited species but made no attempt to quantify the product
branching ratios.
It is harder to match our data with the measurements and
interpretations of Miyoshi et al.4 and Kitchen et al.16 The
latter group performed photofragment translational spectros-
copy experiments on acrylic acid using an ArF excimer laser
the barrier separating HOCO from H ] CO .7
2
Our observations show no evidence of formation of CO
2
(j \ 193 nm) for photolysis and electron impact (200 eV) mass
spectrometry to detect products. Their main observations, and
certainly those of most relevance in a comparison with the
present work, were made at mass-to-charge ratios of m/e \ 44
and m/e \ 17. Each distribution showed two peaks with
maxima at essentially the same Ñight times for m/e \ 44 and
m/e \ 17. On this basis, Kitchen et al. ascribed both peaks at
each m/e to daughter ions of HOCO. To explain the di†erence
in Ñight times, and the derived centre-of-mass product trans-
lational energy distributions, they invoked the production of
an electronically excited state of HOCO at ca. 150 kJ mol~1
from a slow, tunnelling, dissociation of HOCO. There is no
change in the relative yields of CO and HOCO when the
2
total helium pressure is increased from 10 to 50 Torr (see Fig.
3) nor is there any slow rise in the CO signal that can be
2
ascribed to slow decomposition of HOCO. Moreover, we
point out that in the experiments of Miyoshi et al., vibra-
tionally excited HOCO would have to survive something like
105 collisions with H for tunnelling to be responsible for the
2
observed decay, since it takes place over several milliseconds.
We point out that the secondary chemistry will be far more
complex in the case where HOCO is produced from acrylic
acid, rather than from the Cl ] HCOOH reaction, since all
three photochemical channels that we observe, and possibly
other channels (see Table 1), generate radicals. The radicals
will undoubtedly undergo secondary chemistry including
pressure-dependent radicalÈradical reactions. It seems possible
that some of this chemistry, which is not present in the case
where HOCO radicals are produced from Cl ] HCOOH,
may be responsible for the removal of HOCO on long time-
scales observed in the experiments of Miyoshi et al.
excitation energy and they concluded that neither loss of CO
2
Table 2 Branching ratios for di†erent channels in the photo-
dissociation of triÑuoroacetic and acrylic acids
Products
R \ CF
R \ CH xCH
3
2
RCOOH ] R ] HOCO
RCOOH ] R ] H ] CO
0.28 ^ 0.07
0.61 ^ 0.09
0.32 ^ 0.08
0.37 ^ 0.08
2
RCOOH ] RH ] CO
2
RCOOH ] R ] OH ] CO
RCOOH ] ROH ] CO
0.11 ^ 0.06
0.31 ^ 0.09
The present experiments clearly show that the ultraviolet
photodissociation of triÑuoroacetic and acrylic acids proceeds
Phys. Chem. Chem. Phys., 1999, 1, 1447È1454
1453

View Article Online
Williams, J. Chem. Soc., Faraday T rans. 2, 1988, 84, 105; (d) M. J.
Frost, P. Sharkey and I. W. M. Smith, J. Phys. Chem., 1993, 89,
by at least three channels [ those producing HOCO, CO
2
and CO. In the case of the decarboxylation and decarbonyla-
tion, our measurements provide no information as to whether
the co-products are molecular (i.e., RH and ROH) or two rad-
icals (i.e., R ] H and R ] OH). The experiments of Mearns
and Back12 on triÑuoroacetic acid do, however, provide com-
1
2254.
6
7
M. Aoyagi and S. Kato, J. Chem. Phys., 1984, 88, 6409.
(a) G. C. Schatz, M. S. Fitzcharles and L. B. Harding, Faraday
Discuss. Chem. Soc., 1987, 84, 359; (b) K. Kudla, A. G. Koures,
L. B. Harding and G. C. Schatz, J. Chem. Phys., 1992, 96, 7465.
D. Fulle, H. F. Hamann, H. Hippler and J. Troe, J. Chem. Phys.,
8
9
0
pelling evidence for a high yield of R \ CF in that case.
3
1
996, 105, 983.
R. Overend and G. Paraskevopoulos, Chem. Phys. L ett., 1977, 49,
09.
It is intriguing to speculate on whether the decarboxylation
and decarbonylation of triÑuoroacetic and acrylic acids occur
simultaneously with, or sequentially to, the channel (P.1)
which yields HOCO. It is possible that Ðssion of the CwC
bond generates a fraction of the HOCO with sufficient inter-
nal energy to undergo further decomposition. However, the
1
1
R. Forster, M. Frost, D. Fulle, H. F. Hamann, H. Hippler, A.
Schlepegrell and J. Troe, J. Chem. Phys., 1995, 103, 2949.
11 J. Nolte, J. Grussdorf, F. Temps and H. Gg. Wagner, Z. Natur-
forsch., 1993, 48a, 1234.
1
2 A. M. Mearns and R. A. Back, Can. J. Chem., 1963, 41, 1197.
high relative yields of both CO and CO from acrylic acid,
2
13 (a) R. N. Rosenfeld and B. R. Weiner, J. Am. Chem. Soc., 1983,
105, 3485; (b) R. N. Rosenfeld and B. R. Weiner, J. Am. Chem.
Soc., 1983, 105, 6233.
assuming the co-products to be C H ] H and C H ] OH
2
3
2 3
suggest that this is unlikely since the exothermicity of channel
(
P.1) in this case (see Table 1) is only slightly greater than the
14 P. C. Lessard, PhD thesis, University of California, Davis, 1990.
15 H. Morita, K. Fuke and S. Nagakura, Bull. Chem. Soc. Jpn., 1976,
thresholds for decomposition of HOCO to either H ] CO or
2
4
9, 922.
OH ] CO. Consequently, the HOCO formed in (P.1) would
1
6
7
D. C. Kitchen, N. R. Forde and L. J. Butler, J. Phys. Chem. A,
have to absorb essentially the whole of the exothermicity as
internal energy if HOCO were to decompose and the likeli-
hood that a large fraction of the HOCO radicals are formed
within this range of internal energy seems small.
1
997, 101, 6603.
1
G. Guelachvili and K. N. Rao, Handbook of Infrared Standards,
Academic Press, New York, 1986.
18 A. G. Maki and N. K. Wells, W avenumber Calibration T ables
from Heterodyne Frequency Measurements, NIST Special Pub-
lication, NIST/SP821, 1991.
In conclusion, our results have conÐrmed the multichannel
nature of the ultraviolet photodissociation of these two car-
boxylic acids and we have determined the relative yields in
three of these channels. We believe that studies of these pro-
cesses using pulsed lasers to produce and observe products
could provide fascinating insights into the photodissociation
dynamics of these species.
1
9
(a) H. Sauren, A. Winkler and P. Hess, Chem. Phys. L ett., 1995,
39, 313; (b) R. E. Kagarise, J. Chem. Phys., 1957, 27, 519; (c) W.
R. Feairheller, Jr. and J. E. Katon, Spectrochim. Acta, 1967, 23A,
225; (d) S. W. Charles, F. C. Cullen, N. L. Owen and G. A.
Williams, J. Mol. Struct., 1987 157, 17.
20 M. C. Osborne, PhD thesis, University of Birmingham, 1998.
2
2
2
1
F. Lepoutre, G. Louis and J. Taine, J. Chem. Phys., 1979, 70,
2
225.
We are grateful to EPSRC for a research grant and the award
of a studentship (M.C.O). Li Qiang thanks the Royal Society
and the Chinese Academy of Sciences for the award of a
China Royal Fellowship.
2
2
2
3
D. F. Eggers and B. L. Crawford, J. Chem. Phys. 1951, 19, 1554.
J. C. Stephenson, R. E. Wood and C. B. Moore, J. Chem. Phys.,
1
971, 54, 3097.
24 M. C. Osborne, Li Qiang and I. W. M. Smith, to be published.
H. Basch, M. B. Robin and N. A. Keubler, J. Chem. Phys., 1968,
9, 5007.
2
5
4
References
26 R. C. Millikan, J. Chem. Phys., 1963, 38, 2855.
2
2
7
8
D. J. Miller and R. C. Millikan, J. Chem. Phys., 1970, 53, 3384.
M. A. H. Smith, C. P. Rinsland, B. Fridovich and K. N. Rao, in
Molecular Spectroscopy: Modern Research, ed. K. N. Rao, Aca-
demic Press, New York, 1985, ch. 3.
1
(a) H. E. Radford, T. J. Sears and W. Wei, J. Chem. Phys., 1992,
9
7, 3989; (b) T. J. Sears, W. M. Fawzy and P. M. Johnson, J.
Chem. Phys., 1992, 97, 3996.
2
3
4
5
J. T. Petty and C. B. Moore, J. Chem. Phys., 1993, 99, 47; (b) J. T.
Petty and C. B. Moore, J. Mol. Spectrosc., 1993, 161, 149.
J. T. Petty, J. A. Harrison and C. B. Moore, J. Phys. Chem., 1993,
29
30
31
M. F. Arendt, P. W. Browning and L. J. Butler, J. Chem. Phys.,
1
995, 103, 11194.
B. Ruscic, M. Schwartz and J. Berkowitz, J. Chem. Phys., 1989,
9
7, 11194.
A. Miyoshi, H. Matsui and N. Washida, J. Chem. Phys., 1994,
00, 3532.
9
1, 6780.
M. J. Frost, P. Sharkey and I. W. M. Smith, Faraday Discuss.
Chem. Soc., 1991, 91, 305.
1
(a) I. W. M. Smith and R. Zellner, J. Chem. Soc., Faraday T rans.
2
, 1973, 69, 1616; (b) I. W. M. Smith, Chem. Phys. L ett., 1977, 49,
12; (c) J. Brunning, D. W. Derbyshire, I. W. M. Smith and M. D.
1
Paper 8/09064E
1454
Phys. Chem. Chem. Phys., 1999, 1, 1447È1454