Photofragment imaging of HNCO decomposition: Angular anisotropy and correlated distributions

Article abstract of DOI:10.1063/1.473724

Photodissociation of jet-cooled isocyanic acid has been examined by photofragment ion imaging of H(D) from H(D)NCO and CO from HNCO, and by laser induced fluorescence (LIF) of NH(a ¹Δ) from HNCO. Only modest recoil anisotropy is observed in the H+NCO channel at 243.1 nm (β=-0.13±0.05), while the D+NCO channel at approximately the same wavelength reveals no anisotropy (β=0.00±0.05), confirming that the dissociation of H(D)NCO from the opening of the H(D) channel proceeds via vibrational predissociation on the S₀(¹A′) surface. In contrast, substantial anisotropy (β=-0.66±0.08) is observed in the NH(a ¹Δ)+CO channel at 230.1 nm, but this value can correspond to dissociation on either S₀ or S₁. The photolysis region between 243 and 230 nm thus appears important in providing clues to the dissociation mechanism and the competition between different potential energy surfaces. At 217.6 nm, product state distributions exhibit clear dynamical biases. CO is produced in both v=0 and v=1, while NH(a ¹Δ) distributions correlated with different rovibrational levels of CO, although different in shape, are always cold, consistent with the global NH distribution measured by LIF. The NH distributions indicate dissociation on S₁(¹A″), and can be described by Franck-Condon mapping of transition state wave functions in the HNC bending coordinate without additional torque, implying little anisotropy in the potential along that coordinate. On the other hand, a larger torque is manifest in the CO rotational distribution. Although at 217.6 nm the dissociation is likely to be dominated by decomposition on S₁, competition with radiationless decay is still manifest. From analysis .of the CO photofragment velocity distribution at 230.1 nm, the NH(a ¹Δ)+CO dissociation threshold is determined at 42 765 ±25 cm^-1.

Full text of DOI:10.1063/1.473724

Photofragment imaging of HNCO decomposition: Angular anisotropy and correlated
distributions
A. Sanov, Th. Droz-Georget, M. Zyrianov, and H. Reisler
Citation: The Journal of Chemical Physics 106, 7013 (1997); doi: 10.1063/1.473724
View online: http://dx.doi.org/10.1063/1.473724
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/106/17?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Nuclear motion captured by the slow electron velocity imaging technique in the tunnelling predissociation of the
S1 methylamine
J. Chem. Phys. 136, 024306 (2012); 10.1063/1.3675566
Observation of a new phosphorus-containing reactive intermediate: Electronic spectroscopy and excited-state
dynamics of the HPBr free radical
J. Chem. Phys. 123, 144304 (2005); 10.1063/1.2055227
Photodissociation of H I and D I : Polarization of atomic photofragments
J. Chem. Phys. 122, 084301 (2005); 10.1063/1.1850465
High resolution photoabsorption and photofragment fluorescence spectroscopy of water between 10.9 and 12 eV
J. Chem. Phys. 120, 6531 (2004); 10.1063/1.1652566
Determination of velocity distribution, angular distribution, and vector correlation of photofragments using sub-
Doppler fluorescence-imaging method
J. Chem. Phys. 113, 5716 (2000); 10.1063/1.1290470
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:
160.36.178.25 On: Sat, 20 Dec 2014 04:03:17

Photofragment imaging of HNCO decomposition: Angular anisotropy
and correlated distributions
A. Sanov,a) Th. Droz-Georget, M. Zyrianov, and H. Reislerb)
Department of Chemistry, University of Southern California, Los Angeles, California 90089-0482
͑Received 19 November 1996; accepted 22 January 1997͒
Photodissociation of jet-cooled isocyanic acid has been examined by photofragment ion imaging of
H͑D͒ from H͑D͒NCO and CO from HNCO, and by laser induced fluorescence ͑LIF͒ of
1
NH(a ⌬) from HNCO. Only modest recoil anisotropy is observed in the HϩNCO channel at 243.1
nm ͑␤ϭϪ0.13Ϯ0.05͒, while the DϩNCO channel at approximately the same wavelength reveals
no anisotropy ͑␤ϭ0.00Ϯ0.05͒, confirming that the dissociation of H͑D͒NCO from the opening of
the H͑D͒ channel proceeds via vibrational predissociation on the S (1A ) surface. In contrast,
Ј
0
1
substantial anisotropy ͑␤ϭϪ0.66Ϯ0.08͒ is observed in the NH(a ⌬)ϩCO channel at 230.1 nm,
but this value can correspond to dissociation on either S0 or S1 . The photolysis region between 243
and 230 nm thus appears important in providing clues to the dissociation mechanism and the
competition between different potential energy surfaces. At 217.6 nm, product state distributions
1
exhibit clear dynamical biases. CO is produced in both ␯ϭ0 and ␯ϭ1, while NH(a ⌬)
distributions correlated with different rovibrational levels of CO, although different in shape, are
always cold, consistent with the global NH distribution measured by LIF. The NH distributions
indicate dissociation on S (1A ), and can be described by Franck–Condon mapping of transition
Љ
1
state wave functions in the HNC bending coordinate without additional torque, implying little
anisotropy in the potential along that coordinate. On the other hand, a larger torque is manifest in
the CO rotational distribution. Although at 217.6 nm the dissociation is likely to be dominated by
decomposition on S1 , competition with radiationless decay is still manifest. From analysis of the
1
CO photofragment velocity distribution at 230.1 nm, the NH(a ⌬)ϩCO dissociation threshold is
determined at 42 765 Ϯ 25 cmϪ1. © 1997 American Institute of Physics. ͓S0021-9606͑97͒00517-5͔
I. INTRODUCTION
yet fully elucidated. Channel ͑2͒ is known to dominate just
above the opening of channel ͑3͒; however, the importance
of the latter gradually increases and its quantum yield is larg-
est at wavelengths around 200 nm, with channel ͑2͒ being
the second major pathway.1–4,8,10,12,18 The exact branching
ratios and their wavelength dependencies are still not known
with certainty. Above the barrier to dissociation on S1 , com-
petition between decomposition on S1 and predissociation
following radiationless decay to lower states is expected;
moreover, state specific effects following vibrationally medi-
ated photodissociation have recently been observed at exci-
tation energies where dissociation on S1 plays a significant
role.10
The photodissociation of HNCO following excitation to
the S (1A ) state has recently attracted much attention, as it
Љ
1
involves a small molecule which nevertheless can dissociate
via three channels evolving on two or three potential energy
surfaces ͑PES’s͒,1–14
→NH X ⌺Ϫ͒ϩCO X ⌺ϩ͒
͑1͒
͑2͒
͑3͒
3
1
͑
͑
2
2
HNCO→H S͒ϩNCO X ⌸͒
͑
͑
→NH a ⌬͒ϩCO X ⌺ϩ͒.
1
1
͑
͑
In what follows, NH(X ⌺Ϫ) is denoted by 3NH, and
3
1
1
NH(a ⌬) by NH. The origin of the optically bright state
S1 ͑at Ͻ 35 000 cmϪ1) ͑Refs. 15–17͒ lies above the thresh-
old to channel ͑1͒, but below that of channels ͑2͒ and
͑3͒.10,13,14 At low excitation energies above the S1 origin,
intersystem crossing ͑ISC͒ to T1 and internal conversion ͑IC͒
to S0 are possible routes to channels ͑1͒ and ͑2͒. The ener-
getically lowest channel ͑1͒ is spin-forbidden and has only
Several previous studies of HNCO photodissociation
were carried out at photolysis energies relevant to this
work.1–11,13,18,19 In most studies 300 K samples were used,
and some interpretations might have been affected by the use
of bond energies which were later revised.14 Chandler and
1
co-workers determined the energy disposal in NH and CO
following photolysis of 300 K HNCO at several wavelengths
between 230 and 193 nm.1–3 They found cold 1NH rotational
distributions, while the corresponding CO distributions were
hotter and typical of those commonly obtained by the rota-
tional reflection principle. The proposed mechanism was di-
rect dissociation on a repulsive S1 surface. However, the CO
signal was contaminated by a contribution from channel ͑1͒
whose existence was unknown at that time. Very recent ex-
periments of jet-cooled photolysis at 217.6 nm gave a recoil
3
recently been observed by direct detection of NH in 260–
217 nm photolysis.15͑b͒ The competition between the spin-
allowed channels ͑2͒ and ͑3͒ on S0 and S1 , as well as the
relative importance of the spin-forbidden pathway, are not
a͒
Present address: Joint Institute for Laboratory Astrophysics, University of
Colorado, Boulder, Colorado 80309.
b͒
Author to whom correspondence should be addressed; e-mail:
reisler@chem1.usc.edu
J. Chem. Phys. 106 (17), 1 May 1997
0021-9606/97/106(17)/7013/10/$10.00
© 1997 American Institute of Physics
7013
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:
160.36.178.25 On: Sat, 20 Dec 2014 04:03:17

7014
Sanov et al.: Photofragment imaging of HNCO decomposition
anisotropy parameter ␤ϭϪ0.7Ϯ0.2 and showed clearly an
inverted CO͑␯ϭ0͒ rotational distribution, peaking at Jϭ22
9
1
with FWHM ⌬Jϳ15. The NH photofragment yield spec-
trum in jet-cooled samples is structured, but many features
are broad, and an underlying continuum is evident.13
The HϩNCO channel following 193 nm photolysis of
jet-cooled HNCO was studied by Zhang et al. using the high-
Rydberg time-of-flight ͑HRTOF͒ method.7 The NCO frag-
ment was produced with substantial bending excitation ͑al-
though the largest fraction of the available energy was found
in relative translation͒ and ␤ϭϪ0.85 was reported. The dy-
namics was characterized as direct dissociation on a repul-
sive surface. Crim and co-workers, using 300 K HNCO
samples, looked at this channel near its threshold and de-
duced, based on the electronic absorption spectrum,11 a pre-
dissociation mechanism with significant deviations from sta-
tistical behavior in the NCO vibrational distribution; note
however that an incorrect D0(HϩNCO) value was used.
Zyrianov et al. examined the dissociation of jet-cooled
HNCO near the threshold to channel ͑2͒ at ϳ260 nm,13 and
found that NCO was rotationally cold indicating no exit
channel barrier. They suggested that dissociation at this
wavelength was slow, as two-photon absorption competed
effectively with channel ͑2͒ decomposition. Both the absorp-
tion and the NCO photofragment yield spectra exhibited nar-
row spectral features with rotationally resolved bands per-
sisting even above the threshold of channel ͑2͒, and the
linewidths exhibited state-specific effects probably reflecting
the radiationless decay step. They favored a mechanism in-
volving IC followed by predissociation on S0 . In 248 nm
photolysis studies of 300 K HNCO, the angular distribution
was nearly isotropic,8 also in accordance with a predissocia-
tion mechanism.
Thus, despite its small size, HNCO exhibits rich photo-
chemistry involving several PES’s and surface crossings.12,20
Previous results indicate a transition at higher photolysis en-
ergies from a mechanism dominated by radiationless decay
followed by decomposition on lower surfaces ͑i.e., T1 and
S0͒ to a mechanism dominated more and more by dissocia-
tion on S1 . The transition region and the details of the com-
petitive dynamics need to be better understood. Studying the
dynamics over a broad range of photolysis energies also af-
fords examples of vibrational predissociation on PES’s with
either shallow (S1 ,T1) or deep (S0) wells. Evidently, unrav-
eling the complex dynamics of HNCO requires close inter-
play between theory and experiment, with the latter provid-
ing high quality, state-specific data that can serve as
benchmarks for comparisons with theory.
FIG. 1. Schematics of the photofragment ion imaging arrangement ͑not to
scale͒.
imaging technique pioneered by Chandler and Houston,21
which is capable of providing rapidly and efficiently corre-
lated product state distributions, as well as state specific an-
gular distributions.
II. EXPERIMENT
Our ion imaging arrangement, shown schematically in
Fig. 1, is similar to that developed by Chandler and
co-workers.21 In brief, it consists of a single ion-acceleration
stage, a field-free flight-tube and a position-sensitive detec-
tor. A doubly-skimmed ͑1.29 and 0.78 mm diam skimmers͒
pulsed molecular beam containing 2% H͑D͒NCO seeded in 1
atm of He enters the ion acceleration region through a 3 mm
diam hole in the repeller plate. The rotational temperature in
the skimmed beam was not determined directly, but based on
the expansion conditions and previous results it is expected
to be р5 K. The beam velocity vector is aligned along the
ion flight-tube axis. H͑D͒NCO is photolysed with linearly
polarized UV radiation ͑Ϸ0.3–2 mJ/pulse͒ generated by
frequency-doubling the output of an excimer laser ͑Lambda
Physik EMG 101 or 201͒ pumped dye laser ͑Lambda Physik
LPD 3000 or FL 3002͒. The laser beam is focused with a 15
cm focal length lens and intersects the molecular beam at
right angle, halfway between the repeller and extractor, the
latter being a flat stretched piece of 333 lines per inch ͑lpi͒
Ni mesh mounted at the entrance of the flight-tube. The size
of the overlap region of the laser and molecular beam is
Ϸ1.0ϫ0.2ϫ0.2 mm3. H͑D͒ atoms from H͑D͒NCO photodis-
sociation are detected by 2ϩ1 resonance-enhanced multipho-
ton ionization ͑REMPI͒ via the 2 2S←←1 2S transition at
243.1 nm ͑←← denotes a virtual two-photon transition͒,
while CO fragments are interrogated by 2ϩ1 REMPI via the
In this publication we present results obtained by photo-
fragment ion imaging of H and CO from HNCO and D from
DNCO, as well as by laser induced fluorescence ͑LIF͒ of
1NH from HNCO. In particular, we discuss the photodisso-
ciation on HNCO via pathway ͑3͒ at two photolysis energies,
43 455 cmϪ1 ͑230.1 nm͒ and 45 960 cmϪ1 ͑217.6 nm͒ which
lie 690 and 3195 cmϪ1, respectively, above the channel ͑3͒
threshold, and compare the results to dissociation via channel
͑2͒ at 243.1 nm, where only channels ͑1͒ and ͑2͒ are open.
The results demonstrate the power of the photofragment ion
1
1
Q-branches
of
the
B ⌺ϩ←←X ⌺ϩ
or
C ⌺ϩ←←X ⌺ϩ transitions at 230.1 and 217.6 nm, re-
spectively. In all cases the same laser pulse is used for
H͑D͒NCO photolysis and H, D or CO REMPI detection.
Ions are accelerated by the electric field into the 60 cm long
flight-tube terminated by the 40 mm diam microchannel
plate ͑MCP͒ detector ͑Galilleo, Inc.͒. Another 333 lpi Ni
mesh shields the flight-tube from stray electric fields arising
1
1
J. Chem. Phys., Vol. 106, No. 17, 1 May 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:
160.36.178.25 On: Sat, 20 Dec 2014 04:03:17

Sanov et al.: Photofragment imaging of HNCO decomposition
7015
from high voltages applied to the detector. The fine mesh in
front of the detector does not result in imaging of the grid
pattern, as 333 lpi translates into 524 lines of mesh per de-
tector diameter, or Ͼ1 line per CCD camera pixel. The laser
polarization vector is oriented parallel to the plane of the
detector. The output of the MCP particle-multiplier is
coupled to a phosphor screen, and the image generated by
the ions impacting the detector is recorded with a thermo-
electrically cooled CCD camera ͑Princeton Instruments, Inc.;
back-illuminated 512ϫ512 pixel array͒. Signal averaging is
accomplished by accumulating images on the CCD array for
1–2ϫ104 laser firings.
In order to discriminate against products from multipho-
ton dissociation in these one-laser experiments, the experi-
mental conditions were set to discriminate against high-
translational energy products. All the data were modeled
successfully using the known energetics of HNCO decompo-
sition, and no high velocity fragments were evident.
1
The global rotational distribution of NH from HNCO
photodissociation was measured in a separate apparatus22 as
described previously.13 Jet-cooled HNCO was photolyzed at
FIG. 2. ͑a͒ Experimental image of H atoms following 243.1 nm photolysis
of HNCO. ͑b͒ Corresponding image of D atoms from DNCO. In both cases,
the laser polarization is vertical in the figure plane. In ͑c͒ and ͑d͒ 2D cuts
through 3D recoil distributions of H and D products, respectively, recon-
structed by Abel inversion from images ͑a͒ and ͑b͒, respectively, are shown.
Approximate center-of-mass velocity scales are indicated.
1
217.6 nm and NH products were probed by laser-induced
fluorescence ͑LIF͒ under nonsaturated conditions via the
1
1
c ⌸←a ⌬ transition using a second laser. The observed
spectral line intensities were normalized by the rotational
line strength factors to yield JNH-level populations.
The preparation of H͑D͒NCO followed the procedure of
Ashby and Werner,23 which involves the dropwise addition
of a saturated aqueous (H2O or D2O͒ solution of potassium
cyanate ͑KNCO͒ to concentrated phosphoric acid ͑85%
H3PO4 in H2O or D3PO4 in D2O) under vacuum. The
H͑D͒NCO vapor produced in this reaction was condensed in
a liquid nitrogen trap and distilled twice between traps at
Ϫ20 °C and Ϫ78 °C to remove mostly CO2 , NH3 , and
H2O impurities. DNCO experiments were blind to possible
impurities of HNCO since the laser frequency was tuned to
the D atom transition.
the cylindrical symmetry axes of the 3D ion clouds ͑vertical
in the figures͒, as defined by the laser polarization. Clearly,
the H and D atom recoil velocity distributions do not reveal
significant deviations from spherical symmetry. The intensity
distributions peak close to the respective image centers ͓i.e.,
at small H͑D͒ atom velocities͔ indicating significant internal
excitations of the NCO partner fragments. The experimental
2D projections were transformed into the 3D distributions by
means of the Abel inversion as described elsewhere.21,24 Be-
cause of the cylindrical symmetry inherent in the dissocia-
tion when the photolysis laser is linearly polarized, the re-
constructed 3D distributions can be represented by 2D slices
containing the symmetry axes, as shown in Figs. 2͑c͒ and
2͑d͒ for H and D, respectively. The complete 3D distribu-
tions can be obtained by rotating the cuts about their sym-
metry axes. The vertical lines ͑coinciding with the symmetry
axes͒ visible in Figs. 2͑c͒ and 2͑d͒ are caused by noise ac-
cumulated in the process of the inverse Abel
transformation.21 Owing to the small spatial extent of the
overlap between the molecular and laser beams, no deconvo-
III. RESULTS AND ANALYSIS
Shown in Fig. 2͑a͒ is an image obtained by monitoring
the recoil of H atoms from HNCO photodissociation at 243.1
nm ͑41 141 cmϪ1). The observed image is a 2D projection of
the 3D photofragment recoil distribution. In this one-laser
experiment, both HNCO dissociation and H atom detection
were carried out at the same frequency. However, in order to
cover the entire Doppler width of the recoiling H atoms the
laser was scanned over a range of Ϸ3 cmϪ1 while collecting
the image.
lution to account for the finite overlap volume
21͑b͒
͑‘‘deblurring’’͒
was used. The 3D distributions of the H
Figure 2͑b͒ displays an image obtained under similar
conditions by monitoring D atoms from the photolysis of
DNCO at approximately the same wavelength. As a result of
the frequency shift between the H and D transitions, the dif-
ference in excitation energies of HNCO and DNCO in these
one-laser experiments is Ϸ11 cmϪ1. To cover the entire
Doppler width of the D atoms, the laser was scanned over a
and D products were integrated to yield the angular distribu-
tions P(␪), plotted in Figs. 3͑a͒ and 3͑b͒, respectively,
where ␪ is the angle between the recoil velocity and the laser
polarization vector; also shown are fits by the standard recoil
anisotropy function,25
P ␪͒ϰ 2Ϫ␤͒ϩ3␤ cos2 ␪.
͑4͒
͑
͑
range of Ϸ2 cmϪ1
The planes of the images in Figs. 2͑a͒ and 2͑b͒ include
.
For H from HNCO, the best fit is obtained for a recoil an-
isotropy parameter ␤ ϭ Ϫ0.13 Ϯ 0.05. On the other hand, the
J. Chem. Phys., Vol. 106, No. 17, 1 May 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:
160.36.178.25 On: Sat, 20 Dec 2014 04:03:17

7016
Sanov et al.: Photofragment imaging of HNCO decomposition
recoil distribution of D from DNCO is nearly spherically
symmetric with only minor deviations near the cylindrical
symmetry axis (␪ Ϸ 0° and 180°͒. These deviations are arti-
facts resulting from the Abel inversion of the rather noisy
image in Fig. 2͑b͒ and are not significant. Fitting the data in
the angular range ␪ϭ15°–165° gives ␤ϭ0.00Ϯ0.05. For the
perpendicular A ←A transition,16 negative values of ␤ are
Љ
Ј
expected if dissociation is fast on the time scale of parent
rotation, with a limiting value ␤ϭϪ1.0.
At 230.1 nm (43 455 cmϪ1), an image of CO produced
via channel ͑3͒ was recorded by monitoring the
1
1
Q-bandhead of the CO B ⌺ϩ←←X ⌺ϩ transition, whose
wavelength coincides with the photolysis wavelength. As in-
dividual CO rotational lines could not be resolved due to
spectral congestion, a range of low J’s (JCO Ϸ 0–4) was
probed at a fixed laser frequency. The experimental image
shown in Fig. 4͑a͒ was transformed into its 3D recoil distri-
bution as described above, and a 2D cut through the recon-
structed 3D distribution is displayed in Fig. 4͑b͒. The angular
distribution P(␪) and its fit are plotted in Fig. 5͑a͒. The fit
gives ␤ϭϪ0.66Ϯ0.08, and indicates a dissociation lifetime
comparable to the time scale of parent rotation. As the
1
1
Q-branch line strengths of the ⌺ϩ←← ⌺ϩ transitions are
independent of alignment factors,26 the observed CO frag-
ment angular distribution reflects only the effect of the ␮•v
correlation, convoluted with parent lifetime.
Since at ␭ϭ230.1 nm the excitation energy is close to
the opening of channel ͑3͒, the analysis of the product trans-
lational energy distribution P(E) allows an accurate deter-
mination of D0(1NHϩCO). The 3D distribution whose 2D
cut is shown in Fig. 4͑b͒ was integrated to obtain the radial
distribution P(r), where r is the distance from the HN–CO
center of mass in 3D space. As the time allowed for recoil
before product ions arrive at the detector ͑i.e., the ion flight-
FIG. 3. Angular recoil distributions P(␪) ͑open circles͒ of H ͑a͒ and D ͑b͒
from photodissociation of H͑D͒NCO at 243.1 nm and their fits ͑solid lines͒
using the recoil anisotropy function ͓Eq. ͑1͔͒, yielding ␤ϭϪ0.13 and 0.00
for H and D, respectively.
sition. The variation in the photolysis energy due to the spec-
troscopic selection of different CO levels (Jϭ0–30͒ was
Ͻ10 cmϪ1. Figure 6͑a͒ displays an image of CO(␯ ϭ 0,
Jϭ0–4͒, obtained by monitoring the congested
Q-bandhead. Since the vibrational frequencies of CO
time͒ is measured experimentally, P(r) is readily converted
1
to the velocity distribution P( ) and then to the NHϩCO
v
translational energy distribution P(E). Analysis of P(E),
following the procedure described below for the analysis of
the 217.6 nm images, places D0(1NHϩCO) at 42 765Ϯ25
1
1
(C ⌺ϩ) and CO(X ⌺ϩ) are very similar, the
cmϪ1, as shown in Fig. 5͑b͒. Although lower by 75 cmϪ1
,
Q-bandheads of the ͑1-1͒ and ͑0-0͒ C ⌺ϩ←←X ⌺ϩ tran-
1
1
this value is within the error margin of our previous deter-
1
mination of D0(1NHϩCO) based on LIF detection of NH
ϩ10
Ϫ60
13
from jet-cooled HNCO, 42,840
,
and agrees very well
with the 42 710Ϯ100 cmϪ1 value obtained by Brown et al. in
double-resonance photolysis of 300 K samples.10 This new
value is used in all the data analyses reported in this work.
As seen even by visual inspection of Fig. 4 and con-
firmed by analysis of P(E), CO ͑low-J’s) is produced with
1
relatively high average translational energy. Thus, NH cor-
related with low J’s of CO is rotationally cold, in agreement
1
with the global NH distributions obtained in HNCO pho-
tolysis in this energy range,13,15͑b͒ Estimates show that the
1
average NH rotational energy E (NH) correlated with
͗
͘
rot
CO(J Ϸ 0–4) is only 60 cmϪ1, or less than 10% of the total
available energy.
FIG. 4. ͑a͒ Experimental image of CO͑␯ϭ0, JϷ0–4͒ following 230.1 nm
photolysis of HNCO. The actual scale of the image, recorded with a COϩ
flight-time of 25.7 ␮s, is indicated. The laser polarization is vertical in the
figure plane. ͑b͒ 2D cut through the 3D recoil distribution reconstructed by
Abel inversion from image ͑a͒. The center-of-mass velocity scale is indi-
cated.
Figures 6͑a͒–6͑c͒ show representative examples of one-
laser photofragment images obtained near 217.6 nm by
1
1
monitoring the Q-branch of the CO C ⌺ϩ←←X ⌺ϩ tran-
J. Chem. Phys., Vol. 106, No. 17, 1 May 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:
160.36.178.25 On: Sat, 20 Dec 2014 04:03:17

Sanov et al.: Photofragment imaging of HNCO decomposition
7017
FIG. 5. ͑a͒ Angular recoil distribution P(␪) ͑solid line͒ of CO from the
photodissociation of HNCO at 230.1 nm and its fit ͑dashed line͒ using the
recoil anisotropy function ͓Eq. ͑1͔͒ with ␤ϭϪ0.66. ͑b͒ Radial distribution
P(r) of CO from the photodissociation of HNCO at 230.1 nm and its fit
͑dotted line͒ assuming D0(1NHϩCO)ϭ42 765 cmϪ1 ͑see text for details͒;
FIG. 6. ͑Left column͒ Representative experimental images of CO from
HNCO photodissociation at 217.6 nm. ͑a͒ CO͑␯ϭ0, Jр4͒; ͑b͒ CO͑␯ϭ0,
Jϭ15͒ ͑outer ring͒ and CO͑␯ϭ1, Jϭ11͒ ͑inner ring͒; ͑c͒ CO͑␯ϭ0, Jϭ22͒.
The actual scales of the images, recorded with a COϩ flight-time of 13.2 ␮s,
are indicated. In all cases, the laser polarization is vertical in the figure
plane. ͑Right column͒ 2D cuts ͑d͒–͑f͒ through 3D recoil distributions recon-
structed by Abel inversion from images ͑a͒–͑c͒, respectively. The center-of-
mass velocity scales are indicated.
the dashed line shows
a fit with the same parameters but using
D0(1NHϩCO) ϭ 42 840 cmϪ1 ͑Ref. 13͒.
27
sitions are separated by only 1.2 cmϪ1
.
This proximity
leads to spectral overlap in the 2ϩ1 REMPI spectrum; i.e.,
CO͑␯ϭ1͒ is superimposed on CO͑␯ϭ0, Jу10͒. However the
difference of Ϸ2100 cmϪ1 in the translational energies avail-
able to CO ␯ϭ0 and 1 results in a clear spatial separation of
their contributions in the images. Indeed, images recorded in
the range of the spectral overlap exhibit a characteristic
double-ring pattern as seen in Fig. 6͑b͒ for CO͑␯ϭ0,
Jϭ15͒ ͑outer ring͒ superimposed on CO͑␯ϭ1, Jϭ11͒ ͑inner
ring͒. Figure 6͑c͒, in turn, shows an image recorded for
CO͑␯ϭ0, Jϭ22͒, which overlaps the CO͑␯ϭ1, Jϭ19͒ tran-
sition. This ␯ϭ1 rovibrational level is apparently not popu-
lated in the photolysis, and the image exhibits a single ring
tion of images obtained at slightly different wavelengths
within the same JCO indicates that the laser bandwidth is
somewhat narrower than the full width of the CO product
Doppler profiles. Due to the congested nature of the
Q-branch of the CO C ⌺ϩ←←X ⌺ϩ transition, it is im-
possible to scan the laser fully across the Doppler profile of
an individual rotational line without overlapping with neigh-
boring lines. Since our goal was to obtain correlated rota-
tional distributions, the probe laser wavelength was tuned
only around the peak of each rotational transition, neglecting
the wings, in order to minimize overlap. The smaller laser-
exposure of the wings results in preferential detection of
products recoiling at ␪ close to 0° and 180° and distorts the
image anisotropy. We note, however, that the anisotropy in
HNCO dissociation at 217.6 nm was recently measured by
Kawasaki and co-workers who found ␤ϭϪ0.7Ϯ0.2.9 This
value is similar to that obtained by us at 230.1 nm, but one
should exercise caution in comparing these two results as ␤
is very sensitive to parent rotational temperature, which may
1
1
1
pattern again. Note that NH͑␯ϭ1͒ is not accessible at this
photolysis energy.
The images recorded at 217.6 nm ͓Figs. 6͑a͒–6͑c͔͒ ap-
pear less anisotropic than those at 230.1 nm. This observa-
tion is confirmed by inspecting 2D cuts through the corre-
sponding 3D distributions ͑reconstructed by Abel inversion͒
shown in the right column of Fig. 6 ͓panels ͑d͒–͑f͔͒. How-
ever, this is an artifact caused by incomplete integration over
the full extent of the Doppler profile for each JCO . Inspec-
J. Chem. Phys., Vol. 106, No. 17, 1 May 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:
160.36.178.25 On: Sat, 20 Dec 2014 04:03:17

7018
Sanov et al.: Photofragment imaging of HNCO decomposition
be somewhat different in the two experiments.
The left column of Fig. 7 ͓panels ͑a͒–͑c͔͒ displays prod-
uct translational energy distributions P(E) obtained by trans-
forming the radial distributions P(r) associated with the 3D
recoil distributions whose cuts are shown in Figs. 6͑d͒–6͑f͒.
Again, the distribution in Fig. 7͑b͒ is clearly bimodal, appar-
ently due to the fact that two rovibrational states of CO, i.e.,
CO͑␯ϭ0, Jϭ15͒ and CO͑␯ϭ1, Jϭ11͒, are probed simulta-
neously. Additional support for this interpretation ͓as op-
posed to interpreting the bimodal P(E) as a consequence of
1
a bimodal distribution of NH correlated with a single CO
level͔ is found in the observation that the two maxima in
P(E) are indeed separated by Ϸ2000 cmϪ1, i.e., the vibra-
28
1
tional energy spacing in CO(X ⌺ϩ) ͑2143.2 cmϪ1).
It is important to realize that P(E) obtained by the Jaco-
bian transformation of P(r) reflects the convolution of the
product translational energy distribution with a spatial broad-
ening function. Note, for example, that the P(E) plots shown
in Figs. 7͑a͒–7͑c͒ extend beyond the maximum allowed by
energy conservation. These high-energy tails are due entirely
to the limited spatial ͑velocity͒ resolution. However, some
trends regarding the correlated 1NH distributions can be seen
even by visual inspection of the P(E) plots. For example, for
FIG. 7. ͑Left column͒ ͑a͒–͑c͒ Translational energy distributions P(E) cor-
responding to the images in Figs. 6͑a͒–6͑c͒, respectively, obtained in 217.6
nm photolysis of HNCO. Tic marks above the plots indicate the positions of
correlated 1NH rotational levels. In panel ͑b͒ the top comb relates to 1NH
correlated with CO͑␯ϭ0, Jϭ15͒, while the lower comb is for 1NH corre-
lated with CO͑␯ϭ1, Jϭ11͒. ͑Right column͒ Correlated 1NH rotational dis-
tributions ͑d͒–͑f͒ extracted by fitting the state-selected CO radial distribu-
tions obtained from the images in Figs. 6͑a͒–6͑c͒, respectively. Arrows
J
CO ϭ 22 ͓Fig. 7͑c͔͒, P(E) peaks at an energy corresponding
to higher JNH , compared to P(E) for JCOр4 ͓Fig. 7͑a͔͒,
suggesting that 1NH distributions correlated with higher
JCO tend to have less relative population in the lowest JNH
than those correlated with low JCO
.
This observation is confirmed by the correlated 1NH dis-
tributions P(JNH) displayed in Figs. 7͑d͒–7͑f͒, which were
obtained by fitting the state-selected CO radial distributions
mark the highest energetically allowed JNH
.
P(r) ͓or, equivalently, the velocity distributions P( )͔ by
v
linear combinations of the spatially broadened JNH-specific
distributions Pj(r),
shown in Fig. 7͑e͒ for CO͑␯ϭ1, Jϭ11͒. However, all distri-
butions correlated with CO͑␯ϭ1͒ span almost the entire
range of the energetically allowed JNH ͑with a possible ex-
ception of the last allowed level͒.
From the integrated images, we estimate that the CO
␯ϭ1/␯ϭ0 branching ratio in HNCO photolysis at 217.6 nm
is ϳ0.1. Although CO͑␯ϭ1͒ in rotational levels up to
Pmod r͒ϭ
a P r͒,
͑
j j
͑
͚
j
where jϵJNH , Pmod(r) is the resulting fitting function and
1
the array a ϭP(JNH) is the correlated NH distribution.
JCOϭ22 is accessible energetically, we do not observe sig-
͕ ͖
j
For a specific JCO , the positions of the maxima of individual
nificant populations for CO͑␯ϭ1, JϾ16͒.
J
NH-specific distributions Pj(r) are determined by energy
CO photofragment images and correlated 1NH rotational
distributions such as those shown in Figs. 6 and 7͑d͒–7͑f͒,
respectively, were obtained for other CO rovibrational levels
as well, encompassing the full range in which CO͑␯ϭ0, 1͒
populations were significant. CO in ␯ϭ0 was probed at 10
different J’s ranging from 0 to 30 (Jmaxϭ40͒; for CO͑␯ϭ1͒
measurements were performed at 5 different J’s ranging
from 0 to 17 (Jmaxϭ22͒. Figure 8 shows a sum of the corre-
conservation while their shape, resulting from the spatial
broadening, is assumed Gaussian with a width adjusted to fit
the ‘‘fast’’ slope in the experimental distribution P(r).29
Although the extraction of the correlated JNH distribu-
tions is not mathematically unique, sensitivity checks show
that the procedure represents faithfully the general shapes of
1
the distributions ͑see below͒. The NH distributions corre-
1
lated with low J’s of CO͑␯ϭ0͒ have distinct maxima at the
lowest JNH ͓e.g., Figs. 7͑d͒, 7͑e͔͒. They also exhibit a sec-
ond, less intense, feature at higher JNH with a cutoff below
the energetic limit ͓indicated by arrows in Figs. 7͑d͒–7͑f͔͒.
lated NH distributions ͑filled squares͒ weighted as per the
global CO distribution obtained in 217.6 nm HNCO
9
1
photolysis, and compares it to the global NH distribution
͑open circles͒ measured by LIF. Note the nearly quantitative
agreement between the two distributions; both peak at
1
The NH distributions correlated with higher JCO become
somewhat warmer with maxima shifted to JNHϭ3, as illus-
trated in Fig. 7͑f͒ for CO͑␯ϭ0, Jϭ22͒, and JNH populated
almost up to the energetic limit. The correlated NH distri-
JNHϭ2, decrease rather monotonically with increasing JNH
and vanish for JNHϾ10. The small discrepancy can be due to
͑i͒ incomplete selection of probed CO levels included in the
weighted sum; ͑ii͒ a slightly higher detection sensitivity for
1
butions for CO͑␯ϭ1͒ are also cold, peaking at JNH ϭ 2, as
J. Chem. Phys., Vol. 106, No. 17, 1 May 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:
160.36.178.25 On: Sat, 20 Dec 2014 04:03:17

Sanov et al.: Photofragment imaging of HNCO decomposition
7019
The H atom image reveals some structure corresponding
to vibrations of the NCO counter fragment, which is seen
more clearly in the Abel-inverted image ͓Fig. 2͑c͔͒. The cen-
tral peak ͑which accounts for only a small fraction of the
total yield͒ corresponds by energy conservation to the NCO
͑050͒ bending level, while the first ring surrounding it corre-
sponds energetically to the ͑040͒, ͑120͒ or ͑011͒ levels. The
results of Crim et al. suggest that NCO combination bands
are weaker compared to bands of pure bending levels,11 and
thus the ring in the H atom image is more likely to result
from the ͑040͒ level. Although vibrational assignments in
NCO are complicated by the Renner–Teller interactions,33,34
the mere fact that vibrations are partially resolved implies a
modest rotational excitation of NCO. This observation
agrees with previous results obtained in HNCO photolysis at
other excitation energies,7,11 and can be understood in part
based on considerations of angular momentum conservation.
Since in the jet-cooled HNCO rotational excitation is low,
the rotational angular momentum of NCO must be matched
by orbital angular momentum, which cannot be very large
due to the small reduced mass of the HϩNCO system.7 Note
that the structure in the D atom image ͓Figs. 2͑b͒ and 2͑d͔͒ is
less pronounced, possibly reflecting higher rotational excita-
tion of NCO coming from DNCO ͑compared to that from
HNCO͒, as is plausible considering the larger mass of D.
Another possible source of the loss of resolution in the pho-
tolysis of DNCO is a more statistical population of the NCO
vibrational levels.
FIG. 8. ͑Filled squares͒ Sum of 15 correlated 1NH rotational distributions
extracted from 217.6 nm CO photofragment images and weighted as per the
global CO distribution taken at the same wavelength ͑Ref. 9͒. ͑Open circles͒
The global rotational distribution of 1NH in HNCO photodissociation at
217.6 nm measured by LIF.
slow CO in the imaging experiment as a consequence of
incomplete integration over the Doppler profiles of CO tran-
sitions; and ͑iii͒ the nonunique nature of the JNH fits. The
good agreement with the global distribution confirms the va-
lidity of the fitting procedure used here to extract correlated
1NH rotational distributions.
IV. DISCUSSION
1
A. H(D)؉NCO at 243.1 nm
B. NH؉CO at 230.1 nm
Understanding the mechanism of HNCO dissociation is
closely interrelated with discerning the relative roles of S0 ,
S1 , and T1 . It appears that predissociation on S0 is the domi-
nant mechanism of decomposition via channel ͑2͒ from its
threshold at 260.5 nm ͑38 370 cmϪ1) up to at least 243 nm.13
This conclusion is further supported by the lack of significant
recoil anisotropy in the HϩNCO channel at 248 nm ͑Ref. 8͒
and, as reported here, in the H͑D͒ϩNCO channel in disso-
ciation of H͑D͒NCO at 243.1 nm.
Less certain is the mechanism of HNCO decomposition
near the threshold of channel ͑3͒. While HϩNCO products
can in principle be accessed on all three PES’s (S0 , S1 , and
1
T1), NHϩCO can only be produced on S0 and S1 . Since
these surfaces have significant wells, indirect dissociation
occurs in both cases. Thus, the differences in mechanisms
are subtle, and the final outcome will depend on the rate of
IC, the extent of intramolecular vibrational redistribution
͑IVR͒, different densities of states on S0 and S1 and possible
potential barriers. It is expected that above the S1 barrier,
dissociation on this PES will progressively increase in im-
portance, although dissociation on other surfaces will not be
turned off immediately. Calculations performed in a 3D co-
ordinate space indicate barriers on S1 in both the HϩNCO
The small values of the anisotropy parameter suggest
lifetimes of many picoseconds, consistent with RRKM esti-
mates for dissociation on S0 via a reasonably tight TS.30 We
note that a similar trend in ␤ parameters for HNCO and
DNCO was recently observed with 300 K samples at 248
nm.8 There are three potential sources that can contribute to
the lower value of ␤ for DNCO at 243.1 nm. First, although
the photon energies in the two experiments are similar, the
difference in HNCO and DNCO zero-point energies results
1
and NHϩCO channels, but their exact heights are still un-
certain. The former barrier is estimated to be 6000–8000
12,35
cmϪ1, while the latter is under 1000 cmϪ1
.
The longest wavelength at which significant anisotropy
in HNCO photodissociation has been observed so far is
230.1 nm ͓690 cmϪ1 above D0(1NHϩCO)], while 243.1 nm
is the shortest wavelength where the angular distribution is
still rather isotropic. Thus, the energy range between 243 and
230 nm promises to provide clues as to the relative roles of
the different PES’s. At present, it is not known if with 230.1
nm photolysis the S1 barriers either to HϩNCO or to
1NHϩCO are exceeded. The anisotropy parameter ␤ϭϪ0.66
Ϯ0.08 measured at 230.1 nm corresponds, according to the
in available energies of 2760 cmϪ1 and 2250 cmϪ1 , respec-
31
tively, i.e., a difference of ϳ500 cmϪ1
.
Second, the parent
S0 density of states in DNCO is higher than that in HNCO.
Note that the TS frequencies in H–NCO and D–NCO are not
very different, since it is the H͑D͒–N bond that is being
broken. These two considerations result in kHNCO /kDNCO
Ϸ2, as estimated by RRKM calculations.30 Third, the cou-
pling matrix elements involved in the radiationless decay can
differ substantially for HNCO and DNCO.32
J. Chem. Phys., Vol. 106, No. 17, 1 May 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:
160.36.178.25 On: Sat, 20 Dec 2014 04:03:17

7020
Sanov et al.: Photofragment imaging of HNCO decomposition
model of Jonah,36,37 to ␶disϭ1–2 ps ͑for Trotϭ3–10 K͒, if
recoil is axial and deviations from the limiting value ␤ϭϪ1
are due solely to parent lifetime. Although quite fast on the
time scale of parent rotation ͑and thus consistent with disso-
ciation on S1), this rate does not preclude predissociation on
S0 . In fact, the observed anisotropy may reflect the cumula-
tive decomposition rate on S0 via all open channels.38 At this
energy, i.e., 5085 cmϪ1 above D0(HϩNCO), decomposition
via channel ͑2͒ is still the dominant channel,10 and its rate on
S0 can be as fast as few picoseconds, depending on the ge-
ometry and location of the TS.30 Thus, if the IC rate is faster
than the unimolecular reaction rate, substantially negative
values of ␤ can be expected for predissociation on either
following 300 K photolysis at 643, 3035, 5415, and 9529
2
1
cmϪ1 above D0(1NHϩCO). In contrast, the CO(X ⌺ϩ)
rotational distribution measured by Kawasaki and co-
workers is hot and bell-shaped with the maximum at JCO
9
Ϸ22, and E (CO) Ϸ840 cmϪ1
.
Although this estimate
͗
͘
rot
does not take into account the contribution of CO͑␯ϭ1͒ to
the REMPI spectrum, we estimate that the correction due to
the ϳ10% yield of CO͑␯ϭ1͒ does not exceed ϳ0.05 of
E (CO) . Thus, about 0.26 of the total available energy is
͗
͘
rot
partitioned into CO rotation, while the average vibrational
energy, E (CO) ϳ200 cmϪ1, accounts for ϳ0.06 of the
͗
͘
vib
1
total. CO is more internally excited than NH, while excita-
tion of all product internal degrees of freedom accounts for
ϳ0.4 of the total available energy with the rest going into
translation. We note that because of the large mismatch in
NH and CO rotational constants, the angular momentum of
CO in high rotational levels cannot be compensated by NH
rotation to satisfy angular momentum conservation, since the
available energy allows a maximum of JNHϭ13, while the
maximum JCO is 40. Therefore, significant values of L, the
orbital angular momentum, are required to balance the prod-
uct rotations, implying a substantial translational energy re-
lease.
1
S0 or S1 . Note that the very slow increase in the NH yield
compared to the NCO yield near the threshold of channel ͑3͒
is also suggestive of competitive dissociation on the same
PES.13
As with anisotropy, the interpretation of a rotationally
1
cold NH distribution at 230.1 nm is ambiguous. Recall that
at its threshold, 1NH is produced only in its lowest rotational
level and no evidence of a barrier is seen.13 The rotational
1
distribution at 230.1 nm
E
͗
Ϸ60 cmϪ1 for NH corre-
͓
͗
͘
rot
lated with CO(J Ϸ 0–4), or E Ϸ30 cmϪ1 for the global
͘
rot
1NH distribution measured by LIF in this photolysis energy
Although the predominant decomposition mechanism at
217.6 nm is most likely vibrational predissociation on S1 ,
the energy partitioning deviates substantially from PST pre-
dictions, and is dominated by dynamics. This is expected,
since S1 dissociation involves a barrier and a tight TS, and
energy exchange in the exit channel may be incomplete.
Supporting this point, the calculations of Fang et al. show
15͑b͒
region͔
is colder than predicted by phase space theory
͑PST͒.39 This can reflect incomplete energy exchange be-
tween CO and NH rotations due to the large mismatch be-
tween the rotational constants (BNHϭ16.45 cmϪ1, BCO
ϭ1.93 cmϪ1). It may also reflect dynamical constraints on
angular momentum or mapping of bending wave functions
of a tight TS. At 690 cmϪ1 above D0(1NHϩCO), the TS of
a barrierless reaction on S0 may have already tightened con-
siderably, giving rise to rotational distributions which devi-
ate significantly from PST expectations. Measurement of the
1
that the TS on S1 leading to NHϩCO is located at 1.952 Å
along the C–N bond coordinate.20 Franck–Condon consider-
ations indicate that the repulsive slope along the N–C coor-
dinate is accessed in the photoexcitation encouraging domi-
nance of dynamics over statistics. In the absence of
calculations on a realistic PES, we can only compare our
results to relevant limiting models. Since the excess energy
above D0(1NHϩCO) is only ϳ3200 cmϪ1 ͑and the energy
available above the barrier may be even smaller͒, compari-
sons to the Franck–Condon ͑FC͒ model of mapping of TS
wave functions, and to the rotational reflection principle are
appropriate.40–42
1
CO(X ⌺ϩ) rotational distribution from channel ͑3͒ is com-
plicated, since at 230.1 nm its yield from channel ͑1͒ is sig-
nificantly larger;15͑b͒ however, experiments that discriminate
between fast and slow CO may enable the distinction be-
tween the contributions of these two channels.
1
C. NH؉CO at 217.6 nm
1
1
At higher energies, the NHϩCO barrier on S1 is even-
The shapes of the correlated NH distributions are con-
tually exceeded and the excess energy at 217.6 nm ͑3.195
cmϪ1) is above all current estimates of the height of this
barrier.20,35 Thus, decomposition on S1 is expected to com-
pete effectively with IC and ISC. However, the other sur-
faces (S0 , T1) continue to participate, giving rise to a com-
plex interplay between different PES’s; for example, a small
sistent with mapping of the lowest H–N–C bending levels of
the TS. Note that the shape of the wave function at the TS
can be very different from that in the FC region.42 In other
words, even when the wave function in the FC region has a
nodal structure, some or all of the nodes may vanish as it
1
passes through a narrow bottleneck near the TS. For NH
3
LIF signal from NH ͑accessible only via ISC to T1) is still
correlated with CO in the lowest rotational levels, we can
consider CO as an atom with no angular momentum, and
carry out FC mapping calculations as for a triatomic mol-
ecule treating the H–N–C bend as a harmonic oscillator and
assuming that the parent angular momentum is zero.43 Al-
though a gross oversimplication, this model is capable of
observed.15͑b͒
The product state distributions at 217.6 nm exhibit clear
1
dynamical biases. The correlated and global NH distribu-
tions are cold, peaking at the lowest JNH . The average rota-
1
tional energy in NH is E (1NH) Ϸ 215 cmϪ1, or about
͗
͘
rot
1
0.07 of the available energy, as calculated from the global
reproducing the general shapes of the correlated NH distri-
1
distribution. We note that cold NH distributions have also
butions when assuming no exit-channel torque and a ϳ300
been determined in the works of Chandler and co-workers
cmϪ1 bending frequency at the TS ͑about half of the corre-
J. Chem. Phys., Vol. 106, No. 17, 1 May 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:
160.36.178.25 On: Sat, 20 Dec 2014 04:03:17

Sanov et al.: Photofragment imaging of HNCO decomposition
7021
sponding ground state value͒. Such ‘‘softening’’ of the
H–N–C bend is expected from purely geometrical consider-
ations, i.e., the nearly 60% elongation of the C–N bond at
the TS compared to the FC region.20
likely to arise in dissociation on S1 than on S0 , because the
TS in S1 is bent ͑N–C–O angle 120.3°͒,20 while on S0 the
NCO moiety is almost linear.
1
The situation is more complicated when NH is corre-
V. CONCLUSIONS
lated with high JCO ; to the best of our knowledge, the effect
of counter fragment rotation has not yet been treated within
the FC mapping model. If, however, we view the rotational
angular momentum of CO as the counterpart of parent rota-
tion in triatomic dissociation, some insight can be gained
from FC-model calculations carried out for triatomic mol-
ecules at high parent J states within the framework of an
infinite order sudden approximation.44 These calculations
show that parent rotation is partitioned unequally between
product rotation and orbital angular momentum, and results
in a broader range of fragment rotational excitation than in
the case of low parent J.44 Such broadening has also been
observed experimentally in photodissociation of triatomic
molecules which are well described by the FC model when
comparing 300 K to jet-cooled samples.41 By analogy, we
can speculate that counter fragment CO rotation will result in
broader correlated 1NH distributions than that correlated
with CO(J ϭ 0). This is indeed what is observed in our ex-
periments ͓e.g., compare Figs. 7͑d͒–7͑f͔͒. Additional support
for the interpretation that the shapes of 1NH distributions
reflect mapping of H–N–C bending wave functions at the
͑1͒ The decomposition of H͑D͒NCO from the opening of
the H͑D͒ϩNCO channel at 260 nm up to at least 243 nm
proceeds via vibrational predissociation on S0 . This is
supported by the near isotropy of the H and D angular
distributions at 243 nm. At shorter wavelengths, a switch
to a complex interplay between S0 and S1 dissociation
pathways occurs.
͑2͒ Substantial anisotropy is observed in the 1NHϩCO
channel of HNCO at 230.1 nm. The corresponding in-
crease in dissociation rate is consistent with the increase
in rate of a unimolecular reaction on S0 , but direct dis-
sociation on S1 cannot be excluded. Therefore, the in-
crease in anisotropy cannot be unambiguously attributed
to a change in dissociation mechanism.
͑3͒ At 217.6 nm, energy disposal in 1NHϩCO indicates that
the dissociation is dominated by decomposition on S1
͑with some contribution from S0). The available evi-
dence is mostly circumstantial and more experimental
and theoretical work is needed to clarify the details of
the dynamics in this region. Product state distributions
exhibit clear dynamical signatures; rotationally hot CO is
produced in both energetically accessible vibrational lev-
els ͑␯ϭ0 and 1͒, while 1NH correlated with specific rovi-
brational states of CO is always cold. Both the correlated
1
TS is found in the similarity between the NH distributions
correlated with CO͑␯ϭ0͒ and CO͑␯ϭ1͒. Note that for CO͑␯
1
ϭ1͒, NH distributions span almost the entire range of the
energetically allowed J’s, while most of those correlated
with CO͑␯ϭ0͒ cut off below the energetic limit. Thus, the
rotational excitation of 1NH appears to be constrained by
dynamics rather than energetics. The shapes of the H–N–C
bending wave functions at the TS are likely to depend only
weakly on CO vibrational excitation; the latter determines
mostly how many levels can be populated at the TS, i.e., for
1NH correlated with CO͑␯ϭ1͒ the rotational distribution will
reflect mapping only of the lowest level of the TS, while for
1NH correlated with CO͑␯ϭ0͒ some contribution from
higher TS bending levels is also possible.45
Considering interactions beyond the TS, the momentum
of any force exerted on NH during the C–N bond breaking
will be small, since the NH fragment’s center of mass lies
very close to the N atom, almost on the line of force acting
along the C–N bond. Thus, impulsive release will contribute
little to 1NH fragment rotational excitation beyond the extent
of mapping of the H–N–C bending vibration in the TS re-
gion. The small magnitude of the impulsive torque, com-
bined with little anisotropy on S1 in the H–N–C bending
coordinate, can explain why the NH populations peak at low
J’s. A more significant torque in the exit channel is expected
to be exerted by the same force ͑directed along the C–N
bond͒ on CO. This torque will be further modulated by the
anisotropy in the N–C–O bending potential, and this can be
modeled by the rotational reflection principle.41 Such model
reproduced well the CO rotational distributions obtained in
300 K photolysis of HNCO at 193 nm.3͑a͒ As was pointed out
before,3,11 significant torque in the exit channel is more
1
and the global NH distributions can be rationalized in
terms of mappings of TS H–N–C bending wave func-
tions with only modest torque in the exit channel. The
1
differences between NH and CO product distributions
can be understood in terms of post-TS dynamics.
Finally, unraveling the relative roles of S0 , T1 , and S1
in HNCO dissociation and the dynamics on each surface will
evidently require close cooperation between theory and ex-
periment. The results reported in this publication provide
benchmarks for comparisons with theory, and also point to-
wards ranges of photolysis energies where changes in
mechanisms are likely to occur.
ACKNOWLEDGMENTS
We wish to thank David Chandler, Albert Heck, David
Neyer, and Arthur Suits for helpful advice in setting up our
imaging apparatus, and Aaron Katzenstein, a summer partici-
pant in USC’s NSF-REU program, for his help in carrying
out these experiments. We are also grateful to David Chan-
dler for sharing with us the data analysis software, including
the program used for Abel inversion of images. We thank R.
Schinke, K. Morokuma, F. F. Crim, and H.-R. Volpp for
sending us preliminary results and preprints. We benefited
greatly from discussions with F. F. Crim, R. Schinke, S.
Brown, J. Zhang, and C. Wittig. Support by the U.S. Army
Research Office and the National Science Foundation is
J. Chem. Phys., Vol. 106, No. 17, 1 May 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:
160.36.178.25 On: Sat, 20 Dec 2014 04:03:17

7022
Sanov et al.: Photofragment imaging of HNCO decomposition
gratefully acknowledged. Partial funds for the construction
of the imaging setup were obtained from the Department of
Energy.
27 S. M. Koeckhoven, W. J. Buma, and C. A. de Lange, J. Chem. Phys. 99,
5061 ͑1993͒.
28 K. P. Huber and G. Herzberg, in Molecular Spectra and Molecular Struc-
ture ͑Van Nostrand, Princeton, 1979͒, Vol. 4.
29 The rationale behind performing fits in r-space, as opposed to E-space, is
that the major source of experimental broadening is the finite size of the
dissociation/ionization volume. Its size, defined by the intersection of laser
and molecular beams, corresponds to the width of the broadening func-
tion, which is independent of JNH and JCO in r-space.
1 T. A. Spiglanin, R. A. Perry, and D. W. Chandler, J. Phys. Chem. 90,
6184 ͑1986͒.
2 T. A. Spiglanin and D. W. Chandler, J. Chem. Phys. 87, 1577 ͑1987͒.
3
͑a͒ T. A. Spiglanin, R. A. Perry, and D. W. Chandler, J. Chem. Phys. 87,
30 In these estimates, the H–NCO TS frequencies were assumed to include
those of free NCO, i.e., 1290, 2338, and Ϸ500 cmϪ1 ͑C–N and C–O
stretches and NCO bend, respectively͒, and Ϸ100 cmϪ1 for both the in-
plane and out-of-plane H͑D͒–N–C bending vibrations. The parent density
of states was estimated by direct harmonic count. The difference between
HNCO and DNCO was accounted for by replacing the frequency of the
N–H stretch (3538 cmϪ1) by that of N–D (2637 cmϪ1), while leaving
other frequencies unchanged, as the corresponding frequencies of the TS
are expected to change accordingly, canceling out the isotope effect of all
but the N–H͑D͒ stretching vibrations in the RRKM equation.
31 A. L. L. East, C. S. Johnson, and W. D. Allen, J. Chem. Phys. 98, 1299
͑1993͒.
1568 ͑1987͒; ͑b͒ T. A. Spiglanin and D. W. Chandler, Chem. Phys. Lett.
141, 428 ͑1987͒.
4 W. K. Yi and R. Bersohn, Chem. Phys. Lett. 206, 365 ͑1993͒.
5 B. Bohn and F. Stuhl, J. Phys. Chem. 97, 4891 ͑1993͒.
6 B. Ruscic and J. Berkowitz, J. Chem. Phys. 100, 4498 ͑1994͒.
7 J. Zhang, M. Dulligan, and C. Wittig, J. Phys. Chem. 99, 7446 ͑1995͒.
8
͑a͒ R. A. Brownsword, T. Laurent, R. K. Vatsa, H.-R. Volpp, and J.
Wolfrum, Chem. Phys. Lett. 249, 162 ͑1996͒; ͑b͒ 258, 164 ͑1996͒.
9 M. Kawasaki, Y. Sato, K. Suto, Y. Matsumi, and S. H. S. Wilson, Chem.
Phys. Lett. 251, 67 ͑1996͒.
10 S. S. Brown, H. L. Berghout, and F. F. Crim, J. Chem. Phys. 105, 8103
͑1996͒.
11
͑a͒ S. S. Brown, H. L. Berghout, and F. F. Crim, J. Chem. Phys. 102, 8440
32 E. S. Medvedev and V. I. Osherov, in Radiationless Transitions in Poly-
atomic Molecules ͑Springer, Berlin, 1995͒, Vol. 4.
͑1995͒; ͑b͒ S. S. Brown, R. B. Metz, H. L. Berghout, and F. F. Crim, J.
Phys. Chem. 100, 7948 ͑1996͒; ͑c͒ J. Chem. Phys. 105, 6293 ͑1996͒; ͑d͒
S. S. Brown, C. M. Cheatum, D. A. Fitzwater, and F. F. Crim, ibid. 105,
10911 1996͒; ͑e͒ S. S. Brown, H. L. Berghout, and F. F. Crim, ibid. ͑in
press͒.
33 J. T. Hougen, J. Chem. Phys. 36, 519 ͑1961͒.
34 D. Patel-Mistra, D. G. Sauder, and P. J. Dagdigian, J. Chem. Phys. 93,
5448 ͑1990͒.
35 R. Schinke, J. Klossika, H. Floethmann, and K. Yamashita ͑to be pub-
lished͒.
12 A. M. Mebel, A. Luna, M. C. Lin, and K. Morokuma, J. Chem. Phys. 105,
6439 ͑1996͒.
36 C. Jonah, J. Chem. Phys. 55, 1915 ͑1971͒.
13
͑a͒ M. Zyrianov, A. Sanov, T. Droz-Georget, and H. Reisler, R. Soc.
37 S. Yang and R. Bersohn, J. Chem. Phys. 61, 4400 ͑1974͒.
38 If deviations from the limiting value of ␤ϭϪ1 are due solely to parent
lifetime, then these deviations are a measure of the monitored product
appearance time. Therefore, if decomposition via channels ͑2͒ and ͑3͒
occurs on S0 and IC is much faster than the subsequent unimolecular
reaction, then ␤ measured in the CO channel reflects the cumulative rate
of these two channels, k2 ϩ k3 . The same is true if both channels evolve
mainly on S1 . On the other hand, if the IC rate (kIC) is smaller than k2
ϩ k3 and barriers on S1 are not exceeded, then ␤ of CO is essentially a
measure of kIC . Only if channels ͑2͒ and ͑3͒ evolve on different PES’s
will ␤ of CO reflect the rate of channel ͑3͒ alone; in this case the IC step
would effectively uncouple the two rates from each other since the prob-
ability of recrossing from S0 back to S1 is small based on relative densities
of states considerations.
Chem. Faraday Discuss. 102, 263 ͑1995͒; ͑b͒ M. Zyrianov, T. Droz-
Georget, A. Sanov, and H. Reisler, J. Chem. Phys. 105, 8111 ͑1996͒.
14 From Ref. 13͑b͒, D0(3NHϩCO) ϭ 30 150 Ϯ 60 cmϪ1; D0(HϩNCO)
ϩ10
ϭ 38 370 Ϯ 30 cmϪ1; D0(1NHϩCO) ϭ 42 840
cmϪ1. From Ref. 10,
Ϫ60
D0(HϩNCO)р38 320Ϯ140 cmϪ1;D0(1NHϩCO)ϭ42 710Ϯ100 cmϪ1
.
We recommend the following bond enthalpies: D0(3NHϩCO) ϭ 30 150
Ϯ 60 cmϪ1 ͓Ref. 13͑b͔͒; D0͑HϩNCO͒ ϭ 38 370 Ϯ 30 cmϪ1 ͓Ref. 13͑b͔͒;
D0(1NHϩCO) ϭ 42 765Ϯ 25 cmϪ1 ͑thiswork͒.
15
͑a͒ M. Zyrianov, A. Sazonov, R. A. Beaudet, and H. Reisler ͑to be pub-
lished͒; ͑b͒ Th. Droz-Georget, M. Zyrianov, A. Sanov, and H. Reisler,
Ber. Bunsenges. Phys. Chem. ͑in press͒.
16 R. N. Dixon and G. H. Kirby, Trans. Faraday Soc. 64, 2002 ͑1968͒.
17 H. Okabe, J. Chem. Phys. 53, 3507 ͑1970͒; in Photochemistry of Small
Molecules ͑Wiley–Interscience, New York, 1978͒.
39
18 A. Tezaki, S. Okada, and H. Matsui, J. Chem. Phys. 98, 3876 ͑1993͒.
͑a͒ P. Pechukas, J. C. Light, and C. Rankin, J. Chem. Phys. 44, 794
19
͑1966͒; ͑b͒ P. Pechukas and J. C. Light, ibid. 42, 3281 ͑1965͒; ͑c͒ J. C.
Light, Discuss. Faraday Soc. 44, 14 ͑1967͒.
͑a͒ W. S. Drozdoski, A. P. Baronavski, and J. R. McDonald, Chem. Phys.
Lett. 64, 421 ͑1979͒; ͑b͒ G. T. Fujimoto, M. E. Umstead, and M. C. Lin,
Chem. Phys. 65, 197 ͑1982͒.
40 C. X. W. Qian, A. Ogai, J. Brandon, Y. Y. Bai, and H. Reisler, J. Phys.
Chem. 95, 6763 ͑1991͒.
20 W.-H. Fang, X.-Z. You, and Z. Yin, Chem. Phys. Lett. 238, 236 ͑1995͒.
41 R. Schinke, in Photodissociation Dynamics ͑Cambridge University Press,
Cambridge, 1993͒.
21
͑a͒ D. W. Chandler and P. L. Houston, J. Chem. Phys. 87, 1445 ͑1987͒;
͑b͒ A. J. R. Heck and D. W. Chandler, Annu. Rev. Phys. Chem. 46, 335
͑1995͒.
42 H. Reisler, H.-M. Keller, and R. Schinke, Comments At. Mol. Phys. 30,
191 ͑1994͒.
22 A. Ogai, C. X. W. Qian, and H. Reisler, J. Chem. Phys. 93, 1107 ͑1990͒.
23 R. A. Ashby and R. L. Werner, J. Mol. Spectrosc. 18, 184 ͑1986͒.
24 R. N. Bracewell, in The Fourier Transform and its Applications
͑McGraw–Hill, New York, 1986͒.
43 The simulations were carried out following the FC mapping formalism
described in Ref. 40, which is based on J. A. Beswick and W. M. Gelbart,
J. Phys. Chem. 84, 3148 ͑1980͒.
44 H. Grinberg, C. J. Williams, and K. F. Freed, J. Chem. Phys. 100, 9215
͑1994͒.
25
͑a͒ R. N. Zare and D. R. Herschbach, Proc. IEEE 51, 173 ͑1963͒; ͑b͒ R. N.
Zare, Mol. Photochem. 4, 1 ͑1972͒.
26 R. G. Bray and R. M. Hochstrasser, Mol. Phys. 31, 1199 ͑1976͒.
45 A. Sanov, Ph.D. thesis, University of Southern California, 1996.
J. Chem. Phys., Vol. 106, No. 17, 1 May 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:
160.36.178.25 On: Sat, 20 Dec 2014 04:03:17