Energy disposal in CN(X 2Σ+) produced in the 157 nm photodissociation of acrylonitrile

Article abstract of DOI:10.1063/1.1412256

The rovibrational population distribution for the CN(X) fragment produced in the 157.6 nm photodissociation of acrylonitrile was measured. It was observed that the average energies in vibration and rotation of CN were approximately equal and represented about 5% of the available energy with vinyl radical as the co-fragment. It was suggested that dynamical factors played a significant role in energy disposal as the vibrational and rotational distributions were not in good agreement with an energy-conserving prior distribution.

Full text of DOI:10.1063/1.1412256

Energy disposal in CN (X 2 Σ + ) produced in the 157 nm photodissociation of
acrylonitrile
Jingzhong Guo, Tucker Carrington, and S. V. Filseth
Citation: The Journal of Chemical Physics 115, 8411 (2001); doi: 10.1063/1.1412256
View online: http://dx.doi.org/10.1063/1.1412256
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/115/18?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Dynamics of photodissociation of 3,3,3- d 3 -propene at 157 nm: Site effect and hydrogen migration
J. Chem. Phys. 120, 10992 (2004); 10.1063/1.1747812
Photodissociation of CSCl 2 at 235 nm: Kinetic energy distributions and branching ratios of Cl atoms and CSCl
radicals
J. Chem. Phys. 117, 1123 (2002); 10.1063/1.1480272
Branching ratios of C 2 products in the photodissociation of C 2 H at 193 nm
J. Chem. Phys. 114, 9821 (2001); 10.1063/1.1370942
Competing isomeric product channels in the 193 nm photodissociation of 2-chloropropene and in the
unimolecular dissociation of the 2-propenyl radical
J. Chem. Phys. 114, 4505 (2001); 10.1063/1.1345877
Photodissociation of C 2 H at 193 nm: Branching ratios for the formation of C 2 in the X 1 Σ g + , A 1 Π u , and B ′
1 Σ g + states
J. Chem. Phys. 107, 9842 (1997); 10.1063/1.475281
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:
128.42.202.150 On: Sun, 23 Nov 2014 09:02:00

JOURNAL OF CHEMICAL PHYSICS
VOLUME 115, NUMBER 18
8 NOVEMBER 2001
2
Energy disposal in CN„X ⌺¿… produced in the 157 nm photodissociation
of acrylonitrile
Jingzhong Guo, Tucker Carrington, and S. V. Filseth
Department of Chemistry, York University, Toronto, Ontario M3J 1P3 Canada
͑Received 29 May 2001; accepted 27 August 2001͒
Photodissociation of acrylonitrile has been studied at 157.6 nm through analysis of laser-induced
fluorescence experiments. The CN(X 2⌺ϩ) radicals formed in this dissociation are detected in
vibrational levels up to ϭ6 and rotational population distributions are measured for vibrational
v
levels ϭ0–3. The average energies found in vibration and rotation of CN are approximately equal
v
and represent about 5% of the available energy if the vinyl radical is the co-fragment. This is close
to the 7% expected on the basis of the equipartition theorem which suggests that energy disposal is
largely statistical. The vibrational and rotational distributions however are not in good agreement
with an energy-conserving prior distribution suggesting that dynamical factors play a significant role
in energy disposal. © 2001 American Institute of Physics. ͓DOI: 10.1063/1.1412256͔
I. INTRODUCTION
secondary dissociation of energetic primary fragments
formed in reactions ͑3͒ and ͑4͒ must be considered at 157.6
nm.
Considerable experimental and theoretical attention has
recently been devoted to the vacuum ultraviolet photofrag-
mentation of halogen and cyano-substituted ethylenes. These
olefins are small enough that ab initio computations can pro-
vide reliable information on the structures and energetics of
intermediates which is not yet available from experiment. At
the same time the multiplicity of dissociation pathways ac-
cessible at 193 and 157 nm affords an opportunity to com-
pare experimental measurements of branching ratios and
product energy disposal with the prediction of several levels
of quantum and classical theory.
The investigation of cyanoethylene ͑acrylonitrile, vinyl
cyanide͒ is additionally interesting because, ͑a͒ cyano-
substituted olefins are present in the interstellar medium,
and, ͑b͒ a diatomic substituent affords an extra perspective
on the fragmentation dynamics through study of energy dis-
posal into its rotation and vibration. Based upon thermo-
chemistry described below, absorption of light at 157.6 nm
can result in fragmentation through seven primary channels
which are given below with their corresponding values of
⌬E°͑0͒ in kJ/mol,
Photodissociation of acrylonitrile has not been studied
previously at 157.6 nm but there have been several investi-
gations at 193.4 nm and some work with nonlaser sources.
Vacuum ultraviolet absorption begins near 210 nm ͑Ref. 1͒
and bands near 193.4 nm and 157.6 nm have been assigned1
*
*
to mixtures of ␲→␲ and ␴→␴ transitions. Gandini and
Hackett2 studied photodissociation in the long-wavelength
limit of absorption at 213.9 nm. Analyzing final products by
gas–liquid chromatography and mass spectrometry they
found the sum of quantum yields for reactions ͑1͒ and ͑2͒ to
equal 0.8 and the ratio of reactions ͑1͒ to ͑2͒ to equal 1.6. No
evidence was found for reactions ͑3͒–͑6͒ which would be
energetically possible at this wavelength. Nishi et al.3 em-
ployed time-of-flight mass spectrometry and dissociated
acrylonitrile in a molecular beam at 193.4 nm. They interpret
their results in terms of the participation of reactions ͑1͒ and
͑5͒, the latter favored in branching ratio by over 3:1. They
were unable to detect the cyanoacetylene product of reaction
͑2͒ found by Gandini and Hackett to occur with 60% of the
probability of reaction ͑1͒ at 213.9 nm.
Subsequent experiments have failed to confirm the im-
portance of reaction ͑5͒. Fahr and Laufer4 were unable to
detect ultraviolet absorption by C2H3 following VUV flash
photolysis and placed an upper limit of 5% on reaction ͑5͒
for dissociation wavelengths longer than 155 nm.
H2CvCH–CN→HCNϩC2H2
162
175
433
1͒
2͒
3͒
͑
͑
͑
→H2ϩHCwC–CN
˙
→HϩH2CvC–CN
˙
455
524
595
643.
4͒
5͒
6͒
7͒
͑
͑
͑
͑
North and Hall5 have measured Doppler line profiles for
CN(X) fragments from the 193.4 nm photodissociation of
acrylonitrile in a gas cell experiment and find a quantum
yield of 0.003Ϯ0.001 for reaction ͑5͒. Finally, Blank et al.6
in agreement with all other authors except Nishi et al., find
that the radical channel, reaction ͑5͒, represents less than 1%
of the total dissociation yield at 193.4 nm.
→HϩHCvCH–CN
→C2H3ϩCN
˙
˙
→HCwCϩH2CN
→:CH2ϩ:CHCN
The molar energy equivalents of radiation at 157.6 and
193.4 nm are 759 and 618 kJ/mol, respectively. Thus reac-
tions 1–6 are energetically permitted at 193.4 nm and all are
available at 157.6 nm. In addition, CN production through
These observations are consistent with the results for
photodissociation of several vinyl fluorides at 193.4 and
157.6 nm. For example, C–F bond breaking, although ener-
0021-9606/2001/115(18)/8411/7/$18.00
8411
© 2001 American Institute of Physics
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:
128.42.202.150 On: Sun, 23 Nov 2014 09:02:00

8412
J. Chem. Phys., Vol. 115, No. 18, 8 November 2001
Guo, Carrington, and Filseth
getically permitted, was found to be a minor process at 193.4
nm in C2F4,7 ͑CFH͒2,8 and C2F3Cl and CH2CClF.9 No evi-
dence was found at all for this reaction in the CF2CH2
dissociation.8 For two of these molecules, photodissociation
has also been studied at 157.6 nm. For both C2F3Cl10 and
CF2CH2,11 C–F bond breaking remains a minor process at
157.6 nm. Thus we anticipate that CN elimination, found to
be a minor process in acrylonitrile at 193.4 nm,4–6 will be so
as well at 157.6 nm.
In this paper we report the first measurements of energy
disposal within the internal degrees of freedom of CN(X)
produced by photodissociation of acrylonitrile at 157.6 nm.
We compare our results with those of Bird and Donaldson12
for the 193.4 nm dissociation.
0, 0 and 0, 2 bands. The magnitudes of the corrections to the
LIF signals were typically less than 10% and not signifi-
cantly dependent on the value of P2 for different bands. The
maximum correction reached 20% for a 60% change in dye
laser energy.
In our experiments, acrylonitrile ͑Aldrich, 99ϩ%͒ vapor
was taken from a sample of liquid held in a reservoir at 0 °C
and flowed slowly through the gas cell. LIF signals were
acquired with a gated integrator, averaged and analyzed as
described below.
III. DATA ANALYSIS
The extraction of population information from spectral
data can be challenging if extensively overlapped features
are prevalent. In some cases, methods based upon a multi-
variate linear regression procedure can yield excellent results
and we have described such an approach for the determina-
tion of CN(A 2⌸) populations from LIF spectra excited in
the CN(B–A) system.14 This approach to LIF spectra ex-
cited in the CN(B–X) system is less successful since the
band consists of fewer branches and two overlapped pairs
instead of 12 as in the CN(B–A) system. Thus in the present
case we have made estimates of population distributions
based on the relative intensities of nonoverlapped lines and
used these estimates to refine the distributions by comparison
with computed spectra. We have employed for this compari-
son the simulation software, LIFBASE, prepared at SRI In-
ternational by Luque and Crosley.15
II. EXPERIMENT
The apparatus and experimental procedure have been de-
scribed recently13 and only an abbreviated description is
given here. Experiments were carried out in a 27 L aluminum
gas cell. Unpolarized photolysis radiation was provided at
157.6 nm by an excimer laser operating on mixtures of F2
and He at 25 Hz. Product CN(X, ,N) populations were de-
v
termined, in a manner similar to that described previously,
from the intensities of fluorescence in the CN(B–X) system
excited by absorption of tunable dye laser radiation in the
same system between 387 and 460 nm. Laser induced fluo-
rescence ͑LIF͒ spectra were scanned in overlapping sections
and joined together with the aid of repetitive scans over se-
lected lines taken at the beginning and end of single sections.
The probe laser beam was passed through a wedge depolar-
izer and crossed the excimer beam direction at right angle.
LIF was collected with a filtered photomultiplier tube on an
axis normal to the plane of the two laser beams. Several
Since CN(X) is found in this experiment in vibrational
levels up to ϭ6, we have acquired the LIF spectra in three
v
scans to achieve separation in the vibrational levels to some
extent. One of these scans we call ͑0/0͒ corresponding to
excitation in the ⌬ ϭ0 bands with detection of fluorescence
v
complete scans of the CN( ,N) product distributions were
recorded at pressures of 5–6 mTorr with probe delay times
of 0.5–1.5 ␮s.
in the same bands. This spectrum consists of the absorption
spectra of the ͑0,0͒ through ͑5,5͒ bands detected in emission
through a 20 nm wide interference filter ͑␭maxϭ382.5 nm,
FWHM:372.5–392.5 nm͒. A second scan, called ͑0/Ϫ1͒, cor-
v
A number of preliminary experiments were completed
before LIF spectra of the nascent distribution were acquired.
It was found at high excimer laser energies that CN(B) was
produced in the acrylonitrile photolysis. A dependence on the
second power of the excimer energy was established by dis-
persing the CN(B–X) fluorescence through a monochro-
mator and measuring the fluorescence signal while attenuat-
ing the excimer laser beam with fine mesh metal screens of
known transmission. This signal, observed with a second fil-
tered photomultiplier tube, was subsequently employed to
establish the linear dependence of the LIF signal on excimer
laser energy and to correct LIF scans of the CN(X) distribu-
tion for declining excimer laser energy during an experiment.
A small correction to the LIF signals was made to ac-
count for partial transition saturation and for the dependence
of dye laser energy on wavelength as the probe dye laser is
scanned in sections over its gain curve. The dependence of
the LIF signal on probe laser energy was investigated for the
CN(B–X)0,0 and 0,2 bands and the data was fitted with a
saturation function, P1P2E/(1ϩP2E) where E is the dye
laser energy. The saturation parameter P2 is dependent on
the ratio of the Einstein coefficients B/A, and P2 values
were computed for other bands from those measured for the
responds to ⌬ ϭ0 excitation with observation of fluores-
v
cence through
ϭ355.0 nm, FWHM:351.0–360.0 nm͒. This scan excludes
emission from CN(B, ϭ0) permitting more reliable deter-
a
⌬ ϭϪ1 interference filter ͑␭
v
max
v
mination of CN(X, ϭ1) populations. Finally, a third scan,
v
called ͑Ϫ2/0͒, corresponds to excitation in the ⌬ ϭϪ2 se-
v
quence with fluorescence observed through the same filter
employed for the ͑0/0͒ scans. This combination excludes
contributions from excitation of CN(X, ϭ0,1) and permits
v
more reliable determination of populations of CN(X, у2).
v
Two complete experiments for each of these three scans were
acquired and analyzed both separately and as a merged set.
There is no facility in version 1.5 of LIFBASE for mak-
ing quantitative use of residuals so agreement between ex-
perimental and simulated spectra is judged subjectively. In
addition there is no provision for inclusion of fluorescence
filter specifications so that computed spectra represent detec-
tion of all fluorescence rather than that of single bands as
was the case in the present experiments. Thus to the degree
that different bands in a given sequence differ in their
Franck–Condon factors and/or in their transmission through
the filter, corrections must be made. We make the former
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:
128.42.202.150 On: Sun, 23 Nov 2014 09:02:00

J. Chem. Phys., Vol. 115, No. 18, 8 November 2001
Photodissociation of acrylonitrile
8413
TABLE I. Internal energy in CN(X) produced in the 157 mm photodisso-
ciation of acrylonitrile.
Vibrational
level
Rotational
temperature ͑K͒
Rotational
Vibrational
a
b
energy ͑cmϪ1
͒
distribution fractionc
0
1
2
3
4
5
900Ϯ25
2530Ϯ110
2190Ϯ60
2040Ϯ60
͑2000͒
730Ϯ30
1730
2120
0.64
0.24Ϯ0.03
0.09Ϯ0.03
0.024Ϯ0.004
͑0.005͒
1600Ϯ50
͑0.0006͒
aSingle Boltzmann function fit to the nonoverlapped lines in two merged
datasets.
bComputed from the sum of two Boltzmann functions fitted to the nonover-
lapped lines in two merged datasets ( ϭ0,3) or determined from LIFBASE
v
simulations of LIF spectra ( ϭ1,2).
v
cAll levels referenced to ϭ0 through LIFBASE simulations of LIF spectra.
v
in ϭ3 through analysis of the ͑Ϫ2/0͒ scans. In this case, 33
v
nonoverlapped lines in the ͑1,3͒ band were identified in the R
and P branches between Nϭ2 and 53. A comparison of
FIG. 1. Relative rotational population distribution for CN(X, ϭ3) formed
v
in the 157 nm dissociation of acrylonitrile. Open and filled circles are popu-
lations derived from two separate experiments. The dashed line is a single
Boltzmann function fitted to the combined dataset. The solid line is the sum
of two Boltzmann functions fitted to the combined dataset.
single and double-Boltzmann fits to the ϭ3 distribution
v
determined from nonoverlapped lines is given in Fig. 1.
This approach was not successful in the case of the ro-
tational distributions in ϭ1, 2, and 4 where fewer nonover-
v
correction using the Franck–Condon factors of Prasad and
Bernath16 and assess the significance of the latter using mea-
sured filter transmission curves and band linelists.
lapped lines are available: 7, 15, and 0 lines for these levels
in the ͑0/Ϫ1͒, ͑Ϫ2/0͒, and ͑Ϫ2/0͒ scans, respectively. In the
case of vϭ1 and 2, single-Boltzmann fits to the relative
populations inferred from the few nonoverlapped lines were
used as starting points in the LIFBASE simulation of these
spectra. Optimum double-Boltzmann function parameters
were then determined by successive improvements in the
comparison between experimental and simulated spectra. In
the case of vϭ4, which, as described below represents only
1% of the vibrational distribution, only an approximate
single rotational temperature of 2000Ϯ300 K could be ob-
tained because of the dependence of the simulation on both
rotational temperature and vibrational branching ratio. The
same temperature was chosen for vϭ5 in order to determine
an approximate vibrational branching ratio. Finally, the pres-
ence of population in vϭ6 is indicated by the identification
of the ͑4,6͒ P-branch bandhead in the ͑Ϫ2/0͒ scan but no
relative population information is obtainable.
Since we collect LIF without separation of P and R
branches we have examined the absorption spectra of our
two interference filters to determine the efficiency of trans-
mission of both branches in each of the several bands in the
sequences involved. The bands are narrow and overlapped
and the filters are broad and it was found that the variation
from band to band did not exceed the approximately 10%
noise in the LIF signal itself. For example, over the range of
rotational states detected in the ͑0/0͒ and ͑Ϫ2/0͒ scans the
transmission of the P branch varies from 6.4% to 6.8% and
that of the R branch varies from 6.8% to 8.0% for the ͑0,0͒
band. The corresponding variations are from 6.9% to 7.3%
and from 7.3% to 8.0% in the P and R branches of the ͑1,1͒
band. Accordingly, no corrections were made to the LIF
spectra for varying filter transmissions.
The determination of rovibrational population distribu-
tions from the three duplicated LIF scans was accomplished
in several steps. Rotational quantum number assignments in
the LIF spectra were made by comparison of experimental
and computed spectra employing a linelist generated with
literature17 spectroscopic constants for the B–X system. All
rotational distributions are approximately Boltzmann but,
The vibrational branching ratios are obtained from
LIFBASE simulations in comparison with the three sets of
LIF spectra. The population ratio vϭ1/vϭ0, for example,
was obtained from two ͑0/0͒ scans in the following fashion.
A double-Boltzmann function was fitted separately to the
two sets of vϭ0 populations derived from nonoverlapped
lines in the two ͑0/0͒ scans. A double-Boltzmann function
was also determined for the vϭ1 section of the ͑0/Ϫ1͒ scans
which lies between the ͑1,1͒ and ͑2,2͒ bandheads by com-
parison between simulated and experimental LIF spectra us-
ing LIFBASE. The appropriate pair of double-Boltzmann
functions was then employed in an LIFBASE simulation of
each ͑0/0͒ scan and the vϭ1/vϭ0 vibrational population
ratio was adjusted to produce optimum agreement between
experiment and simulation. The ratios, corrected for emis-
sion Franck–Condon factors,16 obtained in this fashion were
0.28Ϯ0.03 and 0.34Ϯ0.03. An average value of 0.31Ϯ0.03
where comparison is possible for ϭ0 and 3, the sum of two
v
Boltzmanns provides a superior fit to the data. The rotational
distribution in ϭ0 was determined from the two ͑0/0͒ scans
v
by identifying 22 nonoverlapped R-branch lines in the N
ϭ0 to 56 interval and correcting their intensities for the val-
18
¨
ues of the Honl–London factors. A double-Boltzmann
function was fitted to this distribution and the fitted param-
eters were used to generate a simulated spectrum within
LIFBASE for comparison to the experimental spectrum. A
similar approach was employed for the rotational distribution
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:
128.42.202.150 On: Sun, 23 Nov 2014 09:02:00

8414
J. Chem. Phys., Vol. 115, No. 18, 8 November 2001
Guo, Carrington, and Filseth
FIG. 2. Experimental and simulated
͑LIFBASE͒ LIF spectra of CN(X,
v
ϭ2–5) formed in the 157 nm disso-
ciation of acrylonitrile and excited and
observed in ⌬ ϭϪ2 and 0 bands, re-
v
spectively. Both spectra are attenuated
by a factor of 3 at wavelengths longer
than the ͑2,4͒ bandhead.
was used to calculate the fractional populations of ϭ0 and
with a fitted double-Boltzmann function for each level. The
average rotational energies for each vibrational level are
given in Table I and are calculated from the double-
Boltzmann functions fit either to the nonoverlapped lines
v
ϭ1 which are given in Table I. A similar procedure was
v
followed to obtain the population ratio of ϭ2/ ϭ1 from
v
v
the two ͑0/Ϫ1͒ scans, and ϭ3/ ϭ2 from the two ͑Ϫ2/0͒
v
v
scans. The fractional populations in levels ϭ4 and 5 were
( ϭ0,3) or to an LIFBASE simulation based upon fewer
v
v
obtained with LIFBASE simulations of the ͑Ϫ2/0͒ scans
where the rotational temperature was fixed at 2000 K and
they are considered very approximate. All ratios are cor-
rected for the relative Franck–Condon factors of the fluores-
cence transition16 to produce a final set of ratios which are
combined to produce the final branching ratios given in Table
I. An LIFBASE simulation employing the uncorrected vibra-
tional ratios and the double-Boltzmann rotational distribu-
tions is compared in Fig. 2 with an experimental ͑Ϫ2/0͒
scan.
nonoverlapped lines ( ϭ1,2). Also given are rotational tem-
peratures based upon fits of single Boltzmann functions to
v
the nonoverlapped lines ( ϭ0,3) or to the simulations (
v
v
ϭ1,2). Table I includes as well the CN(X) vibrational dis-
tribution found in this work. The fractional populations of
levels ϭ0–5 decrease approximately exponentially in the
v
vibrational quantum number as shown in Fig. 4. The slope of
an equally weighted fit of a linear function of the vibrational
quantum number, , to the best determined vibrational frac-
v
tions ( ϭ0–3) is Ϫ1.08Ϯ0.06.
v
IV. RESULTS AND DISCUSSION
A. Experimental distributions
B. Dissociation channels
The energetics associated with seven different dissocia-
tion channels have been given in the introduction and are
illustrated in the energy level diagram of Fig. 5. As was
recently demonstrated by Direcskei–Kovacs and North,19
The rotational distributions for ϭ0 and 3 obtained by
the procedure described above are given in Fig. 3 together
v
FIG. 3. Relative rotational population distributions for CN(X, ϭ0,3)
v
formed in the 157 nm dissociation of acrylonitrile. Open circles and filled
squares represent populations derived from two separate experiments each
FIG. 4. Vibrational population distribution for CN(X) produced in the 157
nm dissociation of acrylonitrile. Open circles are the averages of two ex-
periments for each of three different scans of the distribution. Open squares
are the results of a classical-energy forced quantum oscillator model with an
average vibrational energy of 1050 cmϪ1 as described in the text. Filled
circles are the prior distribution with the vinyl radical treated as a symmetric
top.
for ϭ0 and ϭ3, respectively. Dashed lines are the sum of two Boltzmann
v
v
functions fitted to the combined datasets for each level. The solid line is a
joint conditional prior rotational distribution for ϭ0 as described in the
v
text. The pair of lines for ϭ0 is indicated. All three distributions have been
v
normalized to the same distribution sum.
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:
128.42.202.150 On: Sun, 23 Nov 2014 09:02:00

J. Chem. Phys., Vol. 115, No. 18, 8 November 2001
Photodissociation of acrylonitrile
8415
H2CvC–CN→HϩHCwC–CN
͑8͒
͑9͒
→C2H2ϩCN.
Our calculations find the overall endoergicity of these path-
ways to be 592 and 665 kJ/mol, respectively. Dissociation of
either ␣ or ␤-cyanovinyl radicals would leave 94 kJ/mol
͑7900 cmϪ1͒ of energy available for disposal between the
fragments. Our reported CN internal energy distribution
would then correspond to 28% of that amount, a rather sub-
stantial fraction.
Production of one or more of the cyanovinyl radicals has
been established by Blank et al.6 in the 193.4 nm photodis-
sociation of acrylonitrile through photofragment translational
spectroscopy of the H atom coproduct. Further dissociation
of that radical͑s͒ in the 193.4 nm experiments is not possible
energetically by reaction 9 and the observation that about
half the available energy is found in the kinetic energy of the
H atom produced in reaction ͑3͒ excludes reaction ͑8͒ as
well. At 157.6 nm both subsequent dissociation steps are of
course energetically permitted although the formation of
comparably energetic H atoms in reaction ͑3͒ would prob-
ably exclude secondary dissociation via reaction ͑9͒.
That secondary dissociation of a primary photofragment
of this type can occur is established by the observation by
Lee et al.25 of the FϩC2H2 products of the dissociation of
the ␣-fluorovinyl radical found as a primary fragment in the
193.4 nm dissociation of CH2vCCIF. Similarly, slow com-
ponents in the TOF spectra at m/eϭ19 and 26 are assigned
to secondary dissociation of the fluorovinyl radical in the
157.6 nm photodissociation of vinyl fluoride.26
FIG. 5. Product channels energetically accessible at 193.4 and 157.6 nm in
the dissociation of acrylonitrile. Dissociation energies are those calculated in
this work as described in the text.
high level ab initio calculations on acrylonitrile are possible
with available quantum chemistry software packages. We
have completed a set of such calculations for all available
dissociation channels.
All ab initio calculations were carried out using the
GAUSSIAN 98 program package.20 For geometry optimiza-
tions, we employed the hybrid density functional B3LYP
method, i.e., Becke’s three-parameter exchange functional21
with the correlation functional of Lee, Yang, and Parr,22 and
the 6-311G(d,p) basis set.23 Vibrational frequencies and
therefore, zero-point vibrational energy corrections ͑ZPEC͒,
were calculated at the same level. The energies for all critical
species were reevaluated at the B3LYP/6-311G(d,p) geom-
etry using the high level QCISD͑T͒ method, with full elec-
tron correlation, equipped with the larger 6-311ϩϩG(d,p)
basis set containing both polarization and diffuse functions
for all atoms. All relative energies shown in this paper are
corrected with unscaled zero-point vibrational energies
͑ZPEs͒.
We find direct dissociation of acrylonitrile with the pro-
duction of the vinyl and CN radicals via reaction ͑5͒ to be
endoergic by 524 kJ/mol. This can be compared with a value
of 544 kJ/mol obtained by Derecskei-Kovacs and North19
and with a value of 554 kJ/mol from the NIST database.24
Following photodissociation at 157.6 nm via reaction ͑5͒
there will be 235 kJ/mol ͑19 600 cmϪ1͒ of available energy
for disposal between the two fragments. The distribution
which we report in Table I for CN(X) corresponds to ap-
proximately 1100 cmϪ1 in each of vibration and rotation or,
together, about 11% of the available energy for dissociation
by reaction ͑5͒.
C. Vibrational distribution
It has been proposed by Nishi et al.3 that photoexcitation
1
*
of vinyl cyanide at 193.4 nm produces the A (␲,␲ ) state
Ј
which is rapidly lost by internal conversion to form vibra-
tionally excited ground state accompanied by energy ran-
domization among the vibrational degrees of freedom. Dis-
sociation then follows by molecular and free radical channels
with the branching ratio for reaction ͑5͒ favored over reac-
tions ͑1͒ and ͑2͒ by over 3:1. Subsequent investigations how-
ever find little5,6,12 or no4 CN. Those that do detect CN dis-
agree on whether it is produced through dissociation of
vibrationally excited ground state,5 as proposed by Nishi
et al., or by direct dissociation of the electronically excited
state12 produced initially.
Bird and Donaldson12 find that CN( ϭ1) is produced
v
with a fractional population of 14%, considerably higher
than what would be expected from a prior distribution if the
available energy were distributed statistically as might be
expected for a slow barrier-free dissociation on the ground
state surface. For this reason they suggest that dynamical
factors accompanying dissociation from the excited state are
responsible for CN energy disposal.
Additional channels can lead to CN production by sec-
ondary dissociation of internally excited ␣ or ␤-cyanovinyl
radicals produced in reactions ͑3͒ and ͑4͒, respectively. We
find step ͑3͒ to be endoergic by 433 kJ/mol while Derecskei-
Kovacs and North19 report a value of 454 kJ/mol. There is no
experimental value available for comparison. Production of
CN through reaction ͑3͒ requires that an energetic
␣-cyanovinyl radical undergo subsequent dissociation. Two
channels are evidently possible,
North and Hall5 come to a different conclusion based
upon the good agreement they find between measured Dop-
pler absorption profiles and those simulated on the basis of a
prior distribution. Their observation of a fractional popula-
tion for CN( ϭ1) of 0.12 is consistent with the result of
v
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:
128.42.202.150 On: Sun, 23 Nov 2014 09:02:00

8416
J. Chem. Phys., Vol. 115, No. 18, 8 November 2001
Guo, Carrington, and Filseth
Bird and Donaldson but is still too high to agree with a prior
distribution. Thus while the kinetic energy distribution is
consistent with a statistical distribution of available energy,
the vibrational distribution is not. Blank et al.6 also conclude
that CN is produced with near statistical partitioning of the
available energy on the basis of the agreement they find be-
tween the kinetic energy distribution of the vinyl fragment
and a prior distribution.
increase in rotational energy with vibrational level as we
observe for ϭ0–2. Therefore we consider whether our re-
v
sults are more consistent with a statistical disposal of avail-
able energy. The simplest description of the statistical distri-
bution is the energy-conserving prior distribution.28 When
the prior functions themselves are averaged over fractional
energy disposal they satisfy the equipartition theorem. In the
case of CN formation from acrylonitrile, we would expect to
find 1/14 of the available energy in each of CN vibration and
rotation. For an available energy of 19 600 cmϪ1, this would
In our experiments at 157.6 nm with an available energy
of 19 600 cmϪ1 we have found a vibrational distribution for
which the fractional population decreases approximately ex-
mean equal rotational and vibrational energies of 1400 cmϪ1
.
ponentially in as illustrated in Fig. 4. Also shown in Fig. 4
This can be compared with the average energies we observe
of 1050Ϯ80 cmϪ1 in vibration and 1110Ϯ65 cmϪ1 in rota-
tion. The two average energies are equal within their preci-
sions as required by the equipartition theorem, but somewhat
low.
We also inquire whether the form of the distribution is in
agreement with that expected for the prior distribution. To do
this we evaluate the joint probability density28 at total energy
E given CN vibrational energy Ev and rotational energy Er
and find it to be proportional to (EϪEvϪEr)11. When this is
integrated over a single vibrational level bin width and ex-
pressed in terms of the rotational quantum number N consid-
ered as a continuous variable, we find the conditional prob-
ability density to be proportional to (2Nϩ1)(EϪEvϪEr)12.
v
are the results of a discrete integration5,27 of the continuous
prior probability density for CN fragment vibration, (1
Ϫfv)12, where fv is the fraction of the available energy
found as vibrational energy, which itself is obtained by inte-
grating the normalized prior probability density distribution28
over all other internal degrees of freedom in the products.
The good agreement between experimental and prior distri-
butions observed for the lowest vibrational levels is system-
atically lost for higher levels suggesting that purely statistical
energy disposal is too simple an explanation for our results.
Dynamical models for vibrational energy disposal have
been discussed by many authors and reviewed by Simons29
and Schinke.30 Forces arising from both interfragment and
intrafragment potentials have been considered as sources of
fragment vibrational excitation. An early approach suggested
by Holdy et al.,31 the classical-energy forced quantum oscil-
lator model, has found some success when the available en-
ergy is much larger than the vibrational quantum of energy
and when total fv is low as in the present case. The model
predicts a Poisson distribution in the vibrational quantum
number and average vibrational energy. This distribution is
also shown in Fig. 4 and is seen to provide the best agree-
ment with our results. An impulsive dissociation model, as
proposed by Bird and Donaldson12 for the 193.4 nm disso-
ciation, would predict an angle-dependent competition be-
tween vibrational and rotational energy disposal. In our case
however, as shown in Table I, rotational energy disposal ini-
tially increases with vibrational quantum number.
This distribution was evaluated for vibrational levels ϭ0
v
and 3 and is displayed in Fig. 3 for ϭ0 for comparison to
v
the experimental distribution. Agreement between experi-
ment and the energy-conserving conditional prior distribu-
tion is poor for these levels. Moreover, as would be expected,
the prior rotational distributions cool for higher vibrational
levels in contrast to what is observed in the experiment. As
was the case for the vibrational distribution we again con-
clude that purely statistical energy disposal is inconsistent
with our results.
V. SUMMARY
We have measured the rovibrational population distribu-
tion for the CN(X) fragment produced in the 157.6 nm pho-
todissociation of acrylonitrile. We find that on average the
rotational and vibrational energies of CN(X) are equal at
approximately 1100 cmϪ1 which represents 80% of the val-
ues expected on the basis of the equipartition theorem if the
CN is formed as a primary fragment along with the vinyl
radical. The rotational and vibrational distributions are not in
good agreement with the forms expected on the basis of sta-
tistical energy disposal as described by an energy-conserving
prior distribution.
At 193.4 nm the kinetic energy distribution is statistical
but the vibrational energy distribution is not.5 This suggests a
decoupling between these two degrees of freedom during the
dissociation. These are the conditions under which an in-
trafragment Franck–Condon mapping model29,30 might pro-
vide an interpretation of the origin of vibrational energy dis-
posal. The distribution can be calculated with the Fermi
golden rule when final state interactions are ignored. How-
ever, because of the near equivalence in both the free CN
radical and in the CN in excited acrylonitrile, of both the
vibrational frequency and internuclear separation, almost no
vibrational excitation would be anticipated from this source.
A calculation confirms this expectation, predicting less than
1% vibrational excitation in clear disagreement with our re-
sults.
ACKNOWLEDGMENTS
The authors thank Dr. Robert Eng, Dr. Greg Hall, and
Dr. Simon North for helpful communications. This research
has been supported by the Natural Sciences and Engineering
Research Council of Canada and by York University.
1 P. A. Mullen and M. K. Orloff, Theor. Chim. Acta 23, 278 ͑1971͒.
2 A. Gandini and P. A. Hackett, Can. J. Chem. 56, 2096 ͑1978͒.
3 N. Nishi, H. Shinohara, and I. Hanazaki, J. Chem. Phys. 77, 246 ͑1982͒.
4 A. Fahr and A. H. Laufer, J. Phys. Chem. 96, 4217 ͑1992͒.
D. Rotational distribution
An impulsive dissociation as proposed by Bird and
Donaldson12 at 193.4 nm would not be expected to lead to an
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:
128.42.202.150 On: Sun, 23 Nov 2014 09:02:00

J. Chem. Phys., Vol. 115, No. 18, 8 November 2001
Photodissociation of acrylonitrile
8417
5 S. W. North and G. E. Hall, Chem. Phys. Lett. 263, 148 ͑1996͒.
6 D. A. Blank, A. G. Suits, Y. T. Lee, S. W. North, and G. E. Hall, J. Chem.
Phys. 108, 5784 ͑1998͒.
18 I. Kovacs, Rotational Structure in the Spectra of Diatomic Molecules
͑American Elsevier, New York, 1969͒, p. 125.
19 A. Derecskei-Kovacs and W. W. North, J. Chem. Phys. 110, 2862 ͑1999͒.
20 M. J. Frisch, G. W. Trucks, H. B. Schlegel et al., GAUSSIAN 98, Revision
A.6, Gaussian, Inc., Pittsburgh, PA, 1998.
7 T. K. Minton, P. Felder, R. C. Scoles, and J. R. Huber, Chem. Phys. Lett.
164, 113 ͑1989͒.
8 B. A. Balko, J. Zhang, and Y. T. Lee, J. Phys. Chem. A 101, 6611 ͑1997͒.
9 Y. R. Lee, L. D. Wang, Y. T. Lee, and S. M. Lin, J. Chem. Phys. 113, 5331
͑2000͒; 113, 6107 ͑2000͒.
21 A. D. Becke, J. Chem. Phys. 97, 9173 ͑1992͒.
22 C. Lee, W. Yang, and R. G. Parr, Phys. Rev. B 37, 785 ͑1988͒.
23 R. Krishnan, M. Frisch, and J. A. People, J. Chem. Phys. 72, 4244 ͑1988͒.
24 http://webbook.nist.gov/chemistry
10 J. J. Lin, T. C. Hsu, D. H. Hwang, Y. T. Lee, and X. Yang, J. Chem. Phys.
109, 10719 ͑1998͒.
25 Y. R. Lee, L. D. Wang, Y. T. Lee, and S. M. Lin, J. Chem. Phys. 113, 5331
͑2000͒.
11 J. J. Lin, S. M. Wu, D. H. Hwang, Y. T. Lee, and X. Yang, J. Chem. Phys.
109, 10838 ͑1998͒.
26 K. Sato, S. Tsunashima, T. Takayanagi, G. Fijisawa, and A. Yokoyama,
Chem. Phys. Lett. 242, 401 ͑1995͒.
12 C. A. Bird and D. J. Donaldson, Chem. Phys. Lett. 249, 40 ͑1996͒.
13 J. Guo, R. Eng, T. Carrington, and S. V. Filseth, J. Chem. Phys. 112, 8904
͑2000͒.
27 G. E. Hall ͑private communication͒.
28 J. T. Muckerman, J. Phys. Chem. 93, 179 ͑1989͒.
29 J. P. Simons, Gas Kinetics and Energy Transfer ͑The Chemical Society,
London, 1977͒, p. 58.
14 R. Eng, H. M. Lambert, R. Fei, T. Carrington, and S. V. Filseth, Chem.
Phys. Lett. 261, 651 ͑1996͒.
15 J. Luque and D. R. Crosley, ‘‘LIFBASE: Database and Spectral Simula-
tion Program ͑Version 1.5͒,’’ SRI International Report MP 99-009 ͑1999͒.
16 C. V. V. Prasad and P. F. Bernath, J. Mol. Spectrosc. 156, 327 ͑1992͒.
17 H. Ito, Y. Ozaki, K. Suzuki, T. Kondow, and K. Kuchitsu, J. Mol. Spec-
trosc. 127, 143, 283 ͑1998͒.
30 R. Schinke, Photodissociation Dynamics ͑Cambridge University Press,
Cambridge, 1993͒.
31 K. E. Holdy, L. C. Klotz, and K. R. Wilson, J. Chem. Phys. 52, 4588
͑1970͒.
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:
128.42.202.150 On: Sun, 23 Nov 2014 09:02:00