Photodissociation and photoisomerization pathways of the HNCN free radical

Article abstract of DOI:10.1063/1.1367411

The photodissociation dynamics of the HNCN free radical and its photofragment translational spectroscopy was done to characterize the global CH+N₂ potential energy surface. A reaction mechanism was proposed for the dissociation of HNCN to ground state. Photofragment mass distributions indicated N₂ loss to be the primary dissociation pathway. The study suggests that HNCN radical first isomerizes to a cyclic HCN₂ intermediate. The bond dissociation energy D_o and heat of formation were measured to be nearly 2.8eV and 3.35eV respectively.

Full text of DOI:10.1063/1.1367411

Photodissociation and photoisomerization pathways of the HNCN free radical
Ryan T. Bise, Alexandra A. Hoops, and Daniel M. Neumark
Citation: The Journal of Chemical Physics 114, 9000 (2001); doi: 10.1063/1.1367411
View online: http://dx.doi.org/10.1063/1.1367411
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/114/20?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Photodissociation dynamics of the HCNN radical
J. Chem. Phys. 124, 204303 (2006); 10.1063/1.2196890
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
193-nm photodissociation of acryloyl chloride to probe the unimolecular dissociation of CH 2 CHCO radicals and
CH 2 CCO
J. Chem. Phys. 120, 4223 (2004); 10.1063/1.1644096
C–Cl bond fission, HCl elimination, and secondary radical decomposition in the 193 nm photodissociation of allyl
chloride
J. Chem. Phys. 116, 2763 (2002); 10.1063/1.1433965
Photodissociation dynamics of the allyl radical
J. Chem. Phys. 110, 1450 (1999); 10.1063/1.478020
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:
130.113.86.233 On: Wed, 10 Dec 2014 08:12:33

JOURNAL OF CHEMICAL PHYSICS
VOLUME 114, NUMBER 20
22 MAY 2001
Photodissociation and photoisomerization pathways of the HNCN
free radical
Ryan T. Bise,a) Alexandra A. Hoops, and Daniel M. Neumarkb)
Department of Chemistry, University of California, Berkeley, California 94720
and Chemical Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, California 94720
͑Received 22 January 2001; accepted 5 March 2001͒
The photodissociation spectroscopy and dynamics of the HNCN free radical have been investigated
by fast beam photofragment translational spectroscopy. Predissociative transitions for both the
2
2
2
2
˜
˜
˜
˜
B A
X A band and a higher-energy band system assigned to the C A
X A band were
Љ
Ј
Љ
Љ
observed. Photofragment mass distributions indicate that N2 loss is the primary dissociation
pathway. Translational energy distributions reveal a resolved vibrational structure of the N2
fragment, suggesting that the HNCN radical first isomerizes to a cyclic HCN2 intermediate. A
dissociation mechanism is proposed in which electronically excited HNCN undergoes internal
conversion to the ground state, followed by isomerization to cyclic HCN2 and dissociation through
a tight three-center transition state. The HNCN bond dissociation energy D0 and heat of formation
⌬fH0͑HNCN͒ were determined to be 2.80Ϯ0.03 eV and 3.35Ϯ0.03 eV, respectively. © 2001
American Institute of Physics. ͓DOI: 10.1063/1.1367411͔
I. INTRODUCTION
in which the c-HCN2 adduct isomerizes to the lower-energy
HNCN isomer. This dissociates to spin-allowed H(2S)
Ϫ
In this paper we present a study of the photodissociation
dynamics and spectroscopy of the HNCN radical. This radi-
cal and other HCN2 structural isomers have been proposed as
intermediates in the CHϩN2 reaction, an important reaction
in hydrocarbon combustion because of its possible role in the
formation of ‘‘prompt’’ NO.1–5 Specifically, this reaction
may provide a low-energy pathway for the splitting of mo-
3
˜
ϩNCN(X ͚g ) products, with N and NO formed by subse-
quent reactions of the NCN product.
In an effort to further characterize the global CHϩN2
potential energy surface and assess the importance of the
HNCN radical as a possible intermediate in the CHϩN2 re-
action, we have investigated the photodissociation spectros-
copy and dynamics of the HNCN radical. The work pre-
lecular nitrogen via CH(2
͟
)ϩN2→HCNϩN͑4S͒, the initiat-
sented here represents part of an ongoing study of radicals
ing step of the Fenimore mechanism1,6,7 for prompt NO for-
mation; the N and HCN products are subsequently oxidized
rapidly by O atoms and OH radicals to form nitric oxide.
Although the initial reaction is only endothermic by less than
0.1 eV, it is spin forbidden.
20–22
͑NCN, CNN, HCNN͒
that may play a role in the reac-
tions of N2 in flames leading to NO production.
HNCN is the most stable HCN2 isomer. Previous studies
by Clifford et al.23 indicate that the structural isomer HNCN
lies more than 2.8 eV below CHϩN2 products, compared to
1.13 eV for lin-HCNN21 and 1.02 eV for c-HCN2.11,12 The
HNCN radical was first identified spectroscopically by
Herzberg and Warsop,24 who reported a rotationally resolved
The CHϩN2 reaction has been studied in detail both
experimentally1,3,5,8 and theoretically.2,3,9–18 Early experi-
mental work on this reaction has been reviewed by miller
and Bowman.2 More recent shock tube studies by Dean
et al.5 and Lindackers et al.8 have directly detected the
N(4S) atom and show that the N(4S) atom appearance is
correlated with the removal of CH. Numerous theoretical
studies9–12,15 indicate that the reaction pathway with the
highest probability proceeds through an intermediate doublet
cyclic HCN2(c-HCN2) adduct. This then undergoes intersys-
tem crossing to a quartet surface and finally dissociates to
quartet products N(4S)ϩHCN. However, recent calculations
by Cui et al.12 find the thermal rate constant k(T) to be two
orders of magnitude lower than experimental results due to
the slow ISC rate. This result led to a new mechanism pro-
posed by Lin and co-workers13,19 for N and NO production,
2A
2A electronic absorption (T0ϭ28 994 cmϪ1). They
Ј
Љ
determined the molecular structure for both the ground and
2
˜
excited states and assigned this absorption to the B A
Ј
2
˜
X A band of HNCN. The HNC bond angle for the
Љ
ground state was found to be 116.5° with a nearly linear
NCN backbond. Wu et al.25 probed the B A
X A ori-
Љ
2
2
˜
˜
Ј
gin with laser-induced fluorescence, determining the excited
state lifetime to be 20Ϯ5 ns. Dispersed fluorescence showed
the symmetric stretch (␯2ϭ1140 cmϪ1
)
and the
␦(HN–C–N͒ a mode (␯6Ϸ440 cmϪ1) to be vibronically
Љ
active; the ␯ mode was attributed to vibronic coupling via a
6
Renner–Teller-type interaction. Travis and Herzberg26 re-
ported higher-energy absorption bands at 30 475 and 31 550
cmϪ1 from the flash photolysis of diazomethane, showing
complicated subband structure that shifts upon deuteration.
These higher-energy absorption bands have been observed
using a variety of photolysis sources by Basco and Yee,27
a͒
Current address: Bell Laboratories, Lucent Technologies, 600–700 Moun-
tain Avenue, Murray Hill, New Jersey 07974.
b͒
Author to whom correspondence should be addressed. Electronic mail:
dan@radon.cchem.berkeley.edu
0021-9606/2001/114(20)/9000/12/$18.00
9000
© 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:
130.113.86.233 On: Wed, 10 Dec 2014 08:12:33

J. Chem. Phys., Vol. 114, No. 20, 22 May 2001
Isomerization and dissociation of HNCN
9001
Kroto and co-workers,28 and Mathews et al.29
II. EXPERIMENT
The microwave spectrum of HNCN measured by Yama-
moto and Saito30 showed that the unpaired electron occupies
an orbital perpendicular to the molecular plane of symmetry
and yielded an estimated energy difference of approximately
The fast beam photofragment translational spectrom-
eter34–36 used in these studies is shown in Fig. 1. Vibra-
tionally and rotationally cold neutral radicals are produced
by mass-selectively photodetaching a beam of stable nega-
tive ions. The neutral radicals are then photodissociated by a
second laser and the photofragments are detected directly
with high efficiency.
2
2
˜
˜
1.5 eV between the Renner–Teller X A and A A pair.
Љ
Ј
Photoelectron studies by Clifford et al.23 of the HNCNϪ1
anion yielded the electron affinity of HNCN ͑EAϭ2.622
Ϯ0.005 eV͒ and vibrational frequencies of 1049Ϯ162 and
1879Ϯ106 cmϪ1 for the ␯ and ␯ modes, respectively, in
The ion source configuration for HNCNϪ was the same
as that previously used for NCNϪ.37,38 Argon at a stagnation
pressure of ϳ2 atm is expanded through a pulsed molecular
beam valve into a reservoir containing cyanamide at room
temperature and then through a pulsed electric discharge,
generating mainly CNϪ, NCNϪ, and HNCNϪ ions. Previous
studies20 indicate that this source typically produces anions
with rotational and vibrational temperatures of 50 and 200 K,
respectively. D2NCN was synthesized by repeated washing
of the cyanamide crystals with a methanol-d4 /acetic acid-d4
solution ͑20:1 by volume͒ followed by evaporation of the
solvent. From our ion mass distributions, we estimate the
extent of deuteration to be р40%.
The negative ions generated in the source region are ac-
celerated to 8 keV and separated temporally by a Bakker
time-of-flight ͑TOF͒ mass spectrometer39,40 with a mass
resolution (m/⌬m)Ϸ100. The ion of interest is selectively
photodetached by a pulsed dye laser. To generate vibra-
tionally cold radicals, an excimer pumped dye laser is tuned
to 2.74 eV, which is 118 meV above the detachment thresh-
old based upon the photoelectron spectrum of Clifford
et al.23 Undetached ions are deflected out of the beam path.
In the dissociation region, the fast beam of neutrals are
intersected by a second excimer-pumped dye laser. A frac-
tion of the neutrals absorb and dissociate yielding photofrag-
ments detected directly by either the TOF or TPS ͑time and
position sensing͒ microchannel plate detector assemblies
shown in Fig. 1. An aluminum strip is positioned at the cen-
ter of each detector to prohibit undissociated radicals from
impacting the detector, so that the observed signal is entirely
from recoiling photofragments.
2
3
reasonable agreement with matrix infrared31 and the dis-
persed fluorescence studies.25 Higher-energy photoelectron
studies performed in our laboratory32 have located the A A
2
˜
Ј
2
2
˜
˜
state, finding the X A ϪA A splitting to be 0.7 eV. Un-
like the X A state, which displays a nearly vertical photo-
Љ
Ј
2
˜
Љ
detachment spectrum showing little vibrational excitation,
2
˜
the A A state yields a broad band ͑ϳ1 eV in width͒. No
Ј
regular vibrational structure was observed for this state.
Ab initio calculations of the molecular geometry and vi-
brational frequencies of the HNCN ground state were first
performed by Tao et al.33 and, more recently, by several
other groups.11,12,19,23 The calculations are in reasonable
agreement with the experimental vibrational frequencies
given above. The calculations support an acetylenic structure
showing a shorter HNCϵN bond and longer HN–CN bond
with HNC bond angles between 110.8° to 113.6° and a
nearly linear NCN bond angle of 175°, in good agreement
with the spectroscopic results of Herzberg and Travis.24
The HNCN radical provides an intriguing system for
photodissociation studies, with several relatively low-lying
product channels. Based on this work and accepted literature
values, the heats of reaction ͑at 0 K͒, ⌬rsnH0 , for the five
lowest-energy product channels are as follows:
2
˜
HNCN X A ͒
͑
Љ
h␯
ϩ
g
1
2
˜
˜
2.80Ϯ0.03 eV
͑I͒
N X ͚ ͒ϩCH X
͟
͒
͑
2
͑
——
→
1
ϩ
4
˜
2.87Ϯ0.05 eV
3.45Ϯ0.05 eV
͑II͒
Two types of experiments are performed. First, the spec-
troscopy of the dissociative electronic states is examined by
scanning the dissociation laser and monitoring the total flux
of photofragments arriving at the retractable TOF detector,
located 0.68 m from the dissociation laser. The resulting
photofragment yield ͑PFY͒ spectra is complementary to ab-
sorption and fluorescence measurements. We examined the
HCN X ͚ ͒ϩN S͒
͑
͑
1
ϩ
4
˜
͑III͒
HNC X ͚ ͒ϩN S͒
͑
͑
ϩ
g
1
4
˜
˜
3.54Ϯ0.03 eV ͑IV͒
3.72Ϯ0.04 eV ͑V͒.
N X ͚ ͒ϩCH a ͚͒
͑
͑
2
Ϫ
g
2
3
˜
͑
H S͒ϩNCN X ͚ ͒
͑
photolysis of the HNCN radical from 28 800 to 34 850 cmϪ1
.
Dissociation to the lowest-energy channels ͓͑I͒–͑IV͔͒ re-
quires substantial bond rearrangement and/or intersystem
crossing ͑ISC͒ to quartet surfaces.
The photofragmentation studies performed here show
2
˜
that vibronic levels of both the B A state and the higher-
Ј
energy band system ͑31 200–34 700 cmϪ1͒ assigned to the
C A state are predissociative. Photofragment mass distri-
2
˜
Љ
butions show CHϩN2 to be the primary dissociation channel.
Translational energy distributions display a resolved struc-
ture corresponding to vibrational excitation of the N2 photo-
fragment. The formation of N2 in the photolysis of HNCN
suggests that this radical is a likely intermediate in the
CHϩN2 reaction.
FIG. 1. Fast beam photofragment translational spectrometer. The dotted line
separates the radical production section on the left from the photodissocia-
tion experiment on the right.
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:
130.113.86.233 On: Wed, 10 Dec 2014 08:12:33

9002
J. Chem. Phys., Vol. 114, No. 20, 22 May 2001
Bise, Hoops, and Neumark
The fundamental output of the dye laser with a bandwidth of
0.2 cmϪ1 was used to cover the B A
X A band from
Љ
2
2
˜
˜
Ј
28 800 to 29 850 cmϪ1 while the dye laser output was fre-
quency doubled using KDP-R6G and BBO-B crystals to
cover frequencies from 29 850 to 34 850 cmϪ1 with a laser
bandwidth of 0.3 cmϪ1. Saturation and power-broadening ef-
2
2
˜
˜
fects were observed for the B A
X A band at laser flu-
Љ
Ј
ences Ͼ30 mJ/cm2, and it was therefore scanned at fluences
Ͻ20 mJ/cm2. No noticeable saturation effects were seen for
the higher-energy transitions.
Once the spectroscopy of the dissociative states has been
examined, the dissociation dynamics at selected photolysis
energies are investigated. In this experiment, both photofrag-
ments from a single parent radical are detected in coinci-
dence using a TPS detector based upon the concept devel-
oped by de Bruijn and Los.41 This detection scheme has been
described in detail elsewhere.34,35 The TPS detector records
the positions and difference in arrival time of the two pho-
tofragments from a single dissociation event. This informa-
tion is then used to determine the masses of the fragments,
their relative translational energy, ET , and the scattering
angle ␪ between the relative velocity vector of the polarized
dissociation laser. This coincident detection scheme requires
that photofragment mass ratio be less than 4. The photofrag-
ment mass resolution, m/⌬m, is Ϸ10 and the translational
energy resolution is ⌬ET /ETϭ2.0%. As discussed in a pre-
vious paper,20 the relatively poor mass resolution for the
photofragments is due to the ϳ1 mm diameter of the radical
beam at the TPS detector.
FIG. 2. ͑a͒ Photofragment yield ͑PFY͒ spectrum of the HNCN B 2A
˜
Ј
2A vibronic origin, showing ⌬K ϭϮ1 transitions. K stacks are la-
˜
X
Љ
a
a
beled KaЈϪKЉ . ͑b͒ PFY spectra of the higher-energy vibronic band of the
a
B
2A
X
2A transition. This band is assigned to excitation of the ␦͑HN–
˜
˜
Ј
Љ
C–N͒ a
␯
bend mode.
Љ
6
the origin band. The band is attributed to a parallel vibronic
transition ͑see Sec. IV A͒ with Ka selection rules ⌬Kaϭ0,
Ϯ2,4, . . . . The subband Ka values are labeled using the
2
2
˜
˜
same notation described above for the B A
X A origin
Љ
Ј
transition.
III. RESULTS
A. Photofragment yield spectra
2
1. B A ]X 2A transitions
˜
˜
2. Higher-energy transitions
Ј
Љ
2
2
˜
˜
Higher-energy transition originate near 31 500 cmϪ1 and
form a progression spaced by about 1000 cmϪ1, as shown in
Fig. 3͑a͒, with the most intense feature located near 32 500
cmϪ1. A detailed list of the transition energies and relative
intensities for the HNCN and DNCN photofragment yield
spectra is given in Table I, with the vibrational spacing de-
fined as the energy difference between the intensity maxima
of each vibronic transition. The observed PFY transitions are
consistent with a number of previously reported absorption
bands.26–29 Despite extensive scanning, we were unable to
observe the lowest previously reported member of this pro-
Photofragment yield spectra for the B A
X A band
Љ
Ј
are displayed in Figs. 2͑a͒ and 2͑b͒. Figure 2͑a͒ shows the
photofragment yield spectra between 28 850 to 29 170 cmϪ1
and displays sharp, intense features separated by ϳ40 cmϪ1
,
2
2
˜
˜
consistent with the B A
X A origin transition first ob-
Љ
Ј
served in higher resolution absorption and LIF experiment
by Herzberg and Warsop24 and Wu et al.,25 respectively. The
band displays features characteristic of a perpendicular tran-
sition for a nearly prolate symmetric top: a series of sub-
bands each with a sharp Q branch and weaker P and R
branches. Each Q branch is labeled KЈa –KaЉ , indicating the
excited and ground state quantum numbers KЈa and KaЉ , re-
spectively; the spectrum is dominated by ⌬KaϭϮ1 transi-
tions, as expected for a nearly prolate top. A fit to the rota-
tional contour for the P and R branches of the 2-1 subband
yields a temperature of ϳ60 K for rotation about the b and c
axes. The intensity distribution of the Q branches reveals a
hotter distribution with a characteristic temperature of ϳ200
K. Note that the 2-1 Q branch is about twice as intense as the
0-1 Q branch because twice as many transitions contribute to
it.42
gression near 30 475 cmϪ1 26,29
.
Expanded views of the transitions at 31 550 and 32 550
cmϪ1 are shown in Fig. 3͑b͒. As can be seen in Figs. 3͑a͒ and
3͑b͒, the subband structure of the 31 500 cmϪ1 transition is
remarkably different from the higher-energy transitions; it is
comprised of three main subbands of nearly equal intensity
spaced by ϳ24 cmϪ1 that are denoted X, Y, and Z following
the labeling scheme of Mathews et al.29 Feature X is split
into three main peaks split by nearly 5 and 7 cmϪ1, while Y
shows smaller splittings of 2 and 3 cmϪ1. Feature Z is split
into a doublet with a peak separation of 6 cmϪ1. The ob-
served subband structure for this transition is in excellent
agreement with that observed by Kroto and co-workers28 and
by Mathews et al.29
Figure 2͑b͒ shows a new weak band located ϳ560 cmϪ1
to the blue of the origin band. The most intense features are
spaced by approximately 80 cmϪ1, twice that observed for
the origin band, and display broader rotational contours than
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:
130.113.86.233 On: Wed, 10 Dec 2014 08:12:33

J. Chem. Phys., Vol. 114, No. 20, 22 May 2001
Isomerization and dissociation of HNCN
9003
FIG. 4. Photofragment mass distributions for the fragmentation of HNCN
͑squares͒ and DNCN ͑triangles͒ at 3.9 eV showing the photoproducts to be
CH͑CD͒ϩN2.
vibronic transition increases with photon energy, with values
of 5, 10, and 50 cmϪ1, respectively, for the features at
32 545, 33 524, and 34 525 cmϪ1
.
FIG. 3. ͑a͒ PFY spectra of the HNCN radical from 31 000–35 000 cmϪ1
.
For DNCN ͑not shown͒, transitions were observed near
31 530 and 32 547 cmϪ1 that are strongly red degraded. The
peak positions are similar to those observed for HNCN
showing no real isotope effect. A significant isotope effect is
observed for the subband structure of the 31 530 cmϪ1 band
with peak spacings of 9 and 16 cmϪ1 compared to 24 cmϪ1
for HNCN, indicating that this subband structure is due to
the rotational Ka structure. Only a single red degraded peak
is seen for the 32 547 cmϪ1 transitions. These observations
are again in excellent agreement with the absorption studies
of Kroto and co-workers28 and Mathews et al.29
This band has been assigned to the C 2A
X
2A band. ͑b͒ Expanded
˜
˜
Љ
Љ
views of the 31 550 and 32 550 cmϪ1 vibronic transitions. For the 31 550
cmϪ1 band three major subbands are labeled X, Y, and Z and are spaced by
ϳ24 cmϪ1
.
Higher-energy members of this progression, of which
the 32 550 cmϪ1 band in Fig. 3͑b͒ is an example, are domi-
nated by an intense peak to the red that is strongly red-
degraded. Less intense subbands ͓not shown in Fig. 3͑b͔͒ are
located to the blue and do not display any regular subband
spacing. The FWHM of the most intense subbands of each
B. Photofragment mass distributions
Figure 4 shows the photofragment mass distributions for
HNCN and DNCN for a photon energy of 3.9 eV. Our pho-
tofragment mass resolution is only m/⌬mϷ10, so from the
HNCN distribution alone it is difficult to distinguish between
CHϩN2 products ͑13:28 mass ratio͒ and NϩHCN products
͑14:27͒. Note that these channels are quite close energeti-
cally ͑see the Introduction͒ so one does not expect to be able
to use the product kinetic energy distributions to identify the
mass channel. However, an unambiguous determination re-
sults from a comparison with DNCN photodissociation, for
which both product channels yield a mass ratio of 14:28. The
width of the mass distributions for HNCN is identical to that
for DNCN, indicating that only one mass channel contrib-
utes. In addition, the mass of the light photofragment from
HNCN is one mass unit lower than that from DNCN,
whereas the heavy fragment mass is the same. Hence
CHϩN2 must be the dominant product channel in Fig. 4.
Although the H-loss channel is energetically available for
transitions greater than 3.72 eV, our coincident detection
scheme requires a photofragment mass ratio Ͻ4 and we are
therefore insensitive to the H-loss channel.
TABLE I. Photofragment yield transition energies for HNCN and DNCN.
HNCN
DNCN
Transition Relative
Vibrational Transition Relative
Vibrational
͑cmϪ1
͒
intensity spacing ͑cmϪ1
͒
͑cmϪ1
͒
intensity spacing ͑cmϪ1
͒
31 521
31 526
31 543
31 545
31 547
31 568
31 574
5
5
5
5
5
31 506
31 522
31 531
10
8
6
1036
5
4
32 542
8
1024
31 505
32 538
32 545
32 603
32 620
32 640
32 645
2
6
10
4
3
3
3
989
33 521
33 534
33 601
33 644
4
5
3
2
C. Photofragment translational energy distributions
991
The translational energy P(ET) distributions for the
34 525
1.3
2
2
˜
˜
B A
X A origin band and the higher-energy transitions
Љ
Ј
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:
130.113.86.233 On: Wed, 10 Dec 2014 08:12:33

9004
J. Chem. Phys., Vol. 114, No. 20, 22 May 2001
Bise, Hoops, and Neumark
fine the maximum translational energy, EmT ax , corresponding
to fragments with zero internal energy. The corresponding
values of EmT ax at higher photon energies for HNCN and
DNCN, shown by dashed vertical lines in Figs. 5 and 6,
agree very well with the onset of signal in the P(ET) distri-
butions, strongly supporting our assumption of thermody-
namic significance to ETmax . A very weak signal observed
above EmT ax in some of the distributions is attributed to pho-
toexcitation from vibrationally excited levels of the ground
electronic state.
The structure of the P(ET) distributions reflect the inter-
nal energy distribution of the nascent N2ϩCH(CD) photo-
fragments. The observed structure with an energy separation
of ϳ250 meV can be reasonably assigned to the vibrational
excitation of the N2 photofragment. A further comment on
the internal energy distribution of the photofragments will be
made in Sec. IV B.
The distributions change noticeably with increased pho-
ton energy. For HNCN, the lowest-energy transition displays
two major features peaked at 0.67 and 0.44 eV, respectively,
with the former comprising ϳ85% of the total distribution,
with FWHMϭ150 meV and a slight symmetry to lower
translational energy. The width of the corresponding feature
for DNCN, peaked at 0.73 eV, is slightly broader, FWHM
ϭ170 meV. At higher excitation energies the overall widths
of the P(ET) distributions increase and the structured fea-
tures are considerably broader. The higher-energy distribu-
tions are not as sharply peaked toward high translational en-
ergy, e.g., excitation at 4.035 and 4.157 eV for HNCN yields
P(ET) distributions peaked 0.44 and 0.46 eV below the
maximum translational energy. For all photon energies, the
angular distributions were found to be nearly isotropic.
FIG. 5. Translational energy P(ET) distributions for CHϩN2 photofrag-
ments. Photon energies are given and maximum translational energies ETmax
are indicated with dashed vertical lines. Energetic onsets for the vibrational
states of the N2 fragment are denoted with a comb.
are shown for HNCN and DNCN in Figs. 5 and 6. Distribu-
tions for HNCN and DNCN at the same photon energies are
quite similar. All the distributions display sharp onsets at
high translational energy and reveal structured features with
peak separations ranging from 220 to 260 meV. For HNCN,
the P(ET) distribution from excitation at 3.603 eV displays a
particularly sharp onset at 0.795 eV. A similarly sharp onset
at 0.763 eV is seen in the P(ET) distribution for DNCN
photodissociation at 3.599 eV. We assume these onsets de-
IV. ANALYSIS
A. Photofragment yield spectra
2
1. B A ]X 2A transitions
˜
˜
Ј
Љ
2
2
˜
˜
The B A
X A origin, observed in this work via
Љ
Ј
photofragment yield spectroscopy, Fig. 2͑a͒, has been char-
acterized previously in higher resolution absorption24 and
emission studies.25 In this section, we will therefore focus
our attention toward the new band observed in our PFY
spectra centered ϳ560 cmϪ1 to the blue of the origin ͓Fig.
2͑b͔͒. Wu et al.25 determined the a-axis rotational constants
2
2
˜
˜
for the X A and B A states to be 21.298 and 22.300
Љ
Ј
cmϪ1, respectively. The rotational subband spacing in Fig.
2͑a͒ is nearly 80 cmϪ1, approximately 4A , consistent with
Ј
⌬KaϭϮ2 transitions, suggesting a parallel-type transition
with Ka selection rules ⌬Kaϭ0,Ϯ2,4, . . . . The subband
spacing is larger toward the blue, in agreement with the up-
per state a-axis rotational constant being slightly larger than
that for the ground state. Based upon the Ka subband spacing
and intensity, we assign the largest feature near 29 570 cmϪ1
to overlapped ⌬Kaϭ0 bands with contributions expected
from Kaϭ0,1,2, and 3 branches. The exact positions of the
individual ⌬Kaϭ0 bands cannot be clearly ascertained from
our spectrum. The rotational contours of each subband are
broad, lacking the sharp intense Q branch features observed
FIG. 6. P(ET) distributions for the CDϩN2 from DNCN photodissociation.
Photon energies are given and maximum translational energies EmT ax are
indicated with dashed vertical lines. Energetic onsets for the vibrational
states of the N2 fragment are denoted with a comb.
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:
130.113.86.233 On: Wed, 10 Dec 2014 08:12:33

J. Chem. Phys., Vol. 114, No. 20, 22 May 2001
Isomerization and dissociation of HNCN
9005
for the origin band. The broader rotational contours are con-
sistent with a parallel vibronic transition. While the ⌬Ka
Table II, the predicted EmT ax value the HCNN isomer at a
photon energy of 4.035 eV is 2.91Ϯ0.81 eV.21
ϭ0
bandhead
is
most
intense,
the
⌬Ka
Although the PFY spectra in Fig. 3 correspond to
previously observed spectra, no assignments of the respon-
sible electronic transition were made in the earlier work.
The HNCN radical is isoelectronic with the linear
NCO radical, which has electron configuration
. . . (6␴)2(1␲)4(7␴)2(2␲)3. The HNCN radical is bent,
ϭϮ2 subbands show considerable intensity. The energy
spacing of 560 cmϪ1 from the origin suggests that this band
involves an HN–C–N bend mode and the parallel band
structure indicates that the overall vibronic symmetry of this
band is the same as the ground state, A .
Љ
An analogous parallel band, centered at ϳ440 cmϪ1 to
splitting the ␲ orbitals into a and a orbitals and yielding
Ј
Љ
2
2
˜
˜
. . . (6a )2(7a )2(1a )2(8a )2(9a )2(2a )1 for the mo-
the red of the B A →X A has been observed in dispersed
Ј
Љ
Ј
Ј
Љ
Ј
Ј
Љ
fluorescence measurements by Wu et al.25 As in our spectra,
the ⌬KaϭϮ2 subbands show significant intensity. The au-
thors assign this band to the out of plane bend mode,
2
˜
lecular orbital configuration. The perpendicular B A
Ј
2
˜
X A transition corresponds to promotion of an electron
Љ
from a nonbonding 8a to the 2a lone pair on the
Ј
Љ
␦(HN–C–͒a , of the ground state and suggest that this band
becomes active via a Renner–Teller-type mechanism in
Љ
hydrogen-bonded N atom. The next electronic absorption
should correspond to the promotion of an electron from a
2
2
˜
˜
which the X A and A A bands are contaminated by the
Љ
Ј
strongly bonding ␲-type orbital, 1a or 7a , to the out-of-
Љ
Ј
2
2
˜
˜
B A state, giving the X A states some A character.
Ј
Љ
Ј
2
2
˜
˜
plane lone pair orbital 2a , analogous to the B
͟
X
͟
Љ
These authors argue that the mechanism is similar to the
Renner–Teller interaction described by Bolman and Brown43
to account for the ‘‘forbidden’’ parallel bands in the analo-
transition of the NCO radical.44,45 An electronic transition of
this type is expected to give rise to an extended progression
of the HN–C–N symmetric stretch. Excitation of an electron
2
ϩ
2
˜
˜
gous A ͚ –X
͟
band system of NCO. We propose that a
from the 1a or 7a orbital yields a parallel a-type or per-
Љ
Ј
similar Renner–Teller interaction is responsible for the par-
pendicular c-type transition, respectively.
allel vibronic band centered about 560 cmϪ1 to the blue of
The vibronic transitions near 32 500, 33 500, and 34 500
˜
˜
the B state origin in which the zero-point vibrational level of
the X A state is mixed with vibrational states of the B A
state with a character. The parallel band is accordingly as-
signed to the out of plane bend, ␦(HN–C–͒, an a mode of
cmϪ1 clearly do not exhibit the sharp, well-separated Ka
2
2
˜
Љ
Љ
2
˜
structure associated with the perpendicular c-type B A
Ј
Љ
2
˜
X A transition and are instead dominated by an intense
Љ
Љ
subband that looks like a red-degraded R-bandhead ͓see Fig.
3͑b͔͒. The shape of this feature is consistent with a parallel
electronic transition in which the a-axis rotational constants
are similar so that the ⌬Kaϭ0 bands are superimposed on
each other. In addition, the vibrational progression frequency
of ϳ1000 cmϪ1 is similar to the 1050 cmϪ1 symmetric
2
˜
the B A state with an estimated frequency of 560Ϯ10
Ј
cmϪ1. An alternative mechanism for this parallel vibronic
transition would involve vibronic coupling between the vi-
brational levels of the B A state with the nearby C A
2
2
˜
˜
Ј
Љ
state to which the higher-energy bands between 31 500–
34 500 cmϪ1 are assigned below.
Ϫ
u
Ϫ
g
3
3
˜
˜
stretch progression of the B
͚
X
͚
band of the NCN
radical,20,28 and is in reasonable agreement with experimen-
tal values of 1140 cmϪ1 ͑Ref. 25͒ and 1146 cmϪ1 ͑Ref. 31͒
for the HN–C–N ground state symmetric stretch. Hence it is
reasonable to assume that the observed progression is due to
the upper state symmetric stretch. These observations sug-
gest that the transition involves electronic excitation from the
2
2. Higher-energy bands, C A ]X 2A
˜
˜
Љ
Љ
The PFY spectra in Fig. 3 reveals transitions at higher
photon energies starting at 31 500 cmϪ1. The observed tran-
sitions are in excellent agreement with earlier reported ab-
sorption spectra,26–29 in which these bands appeared simul-
2
2
˜
˜
taneously with the rotationally resolved B A
X A band
Љ
Ј
1a orbital to the 2a orbital and it is assigned to the
Љ
Љ
of HNCN. This led Kroto and co-workers28 and Basco and
Yee27 to associate these bands with another excited elec-
tronic state of HNCN. However, based upon isotopic substi-
tution data and partial rotational analysis of the 31 500 cmϪ1
band, Mathews et al.29 assigned the carrier of these bands
near 30 500, 31 500, and 32 500 cmϪ1 to the HCNN species,
an assignment we believe is incorrect. We confidently assign
this transition and the higher-energy transitions to the HNCN
radical based upon two observations. First, a photoelectron
spectrum taken in a separate apparatus in our laboratory us-
ing the same source conditions shows that only the HNCN
isomer is produced upon photodetachment.32 Second, the ex-
perimental thresholds for P(ET) distributions for the bands
between 31 500–34 500 cmϪ1 match the expected ETmax val-
ues determined from the P(ET) distribution of the well-
2
2
˜
˜
C A
X A band.
Љ
Љ
The band structure associated with the 31 550 cmϪ1
band is particularly complex, as demonstrated by the previ-
ous misassignment of this band to the HNCN isomer.29
Based upon the vibrational spacing, this transition appears to
2
2
˜
˜
be a lower vibronic transition of the C A
X A band.
Љ
Љ
However, the regular subband spacing of ϳ24 cmϪ1 is dra-
matically different from the band structure of the higher-
energy transitions and is not consistent with either parallel or
perpendicular electronic transitions. A complicated subband
structure that shifts upon deuteration has also been observed
for a lower member of this progression at 30 450 cmϪ1 26,29
,
this feature was not seen in our PFY spectrum, presumably
because of its low quantum yield for dissociation. The band
2
2
˜
˜
2
2
˜
˜
structure of the lower members of the C A
X A band
Љ
Љ
known B A
X A band of HNCN. These higher-energy
Љ
bands must therefore also correspond to the HNCN isomer.
For a comparison, based upon the heats of formation listed in
Ј
near 30 450 and 31 550 cmϪ1 may be due to perturbations by
the nearby B A state, similar to the strong perturbations
2
˜
Ј
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:
130.113.86.233 On: Wed, 10 Dec 2014 08:12:33

9006
J. Chem. Phys., Vol. 114, No. 20, 22 May 2001
Bise, Hoops, and Neumark
TABLE II. Heats of formation of relevant HxCyNz species.
TABLE III. N2 photofragment vibrational distributions and average vibra-
tional energy for the HNCN and DNCN P(ET) distributions.
Species
⌬fH0 ͑eV͒
⌬fH298 ͑eV͒
7.428Ϯ0.005
Reference
Photon
energy
C
7.3708Ϯ0.005
47
EmT ax
E
͘
vib
Parent
molecule
͗
H
N
CH
NH
N2
CN
CNN
NCN
HCN
HNC
HCNN
HNCN
2.239Ϯ0.000 06 2.2594Ϯ0.000 06 47
͑eV͒
͑eV͒
(␯ϭ0) ͑␯ϭ1͒ ͑␯ϭ2͒ ͑␯ϭ3͒ ͑eV͒
4.8797Ϯ0.001
6.149Ϯ0.013
3.90Ϯ0.17
0
4.8989Ϯ0.001
6.1846Ϯ0.013
3.90Ϯ0.17
0
47
48
47
HNCN
3.6027
3.911
4.035
4.157
3.599
3.909
4.035
0.795
1.115
1.235
1.36
0.764
1.074
1.200
85
43
31
18
83
44
33
17.5
15
39
41
39
17
41
42
33.8
¯
18
25
30
¯
15
24
31.0
¯
0.043
0.216
0.288
0.396
0.049
0.203
0.276
Ref. state
3
13
¯
¯
2
4.513Ϯ0.021
6.16Ϯ0.03
4.83Ϯ0.03
1.34Ϯ0.03
1.92Ϯ0.04
5.02Ϯ0.18
3.35Ϯ0.03
4.498Ϯ0.021
6.15Ϯ0.03
4.82Ϯ0.03
1.336Ϯ0.03
1.95Ϯ0.04
4.98Ϯ0.18
3.31Ϯ0.03
3.03Ϯ0.02
1.38Ϯ0.11
49
21
20
50
51
21
DNCN
Franck–Condon mapping of TS2
17.7 0.427
This work
21
This work with Ref. 23
H2CNN 3.09Ϯ0.20
H2NCN 1.46Ϯ0.11
ture of the P(ET) distributions for HNCN and DNCN at
similar photon energies indicates that the observed structure
is not associated with the C–H vibration and is instead due to
vibration of the N2 photofragment. Furthermore, the spacing
between peak maxima ranges from 220–260 meV, in reason-
able agreement with the N2 vibrational frequency, 292 meV,
compared to 354 meV for CH. The combs shown above the
P(ET) distributions indicate the calculated onsets for vibra-
tionally excited levels of the N2 fragment. The N2 vibrational
distribution for each of the HNCN and DNCN P(ET) distri-
butions are estimated by fitting the vibrational features of the
P(ET) distributions with nearly Gaussian functions with a
slightly asymmetric tail to lower translational energies. The
resulting vibrational distributions and average vibrational en-
2
˜
that have been observed for the corresponding B
͟ and
states of NCO.44–46
2
ϩ
˜
A
͚
B. Translational energy distributions
The P(ET) distributions reveal how available energy is
distributed among the photofragments. The energy balance
for dissociation of HNCN to CHϩN2 is described by Eq. ͑1͒.
h␯ϩEINT HNCN͒
͑
ϭD0 HNCN͒ϩE ϩE N ͒ϩE CH͒,
͑1͒
͑
͑
VR
͑
VR
T
2
ergy, E
are listed in Table III. Table III and Figs. 5 and
͘
͗
vib
where h␯ is the photon energy, EINT(HNCN) is the average
rotational energy of the parent radical about the b and c axes
͑Ϸ60 K͒ plus the rotational energy associated with a particu-
6 show that the N2 vibrational excitation increases as a func-
tion of the overall photon energy. Increasing the photon en-
˜
˜
ergy from the B state origin at 3.603 eV to the C state at
3.911 eV shifts the ␯ϭ0:␯ϭ1 vibrational population ratio
from 6:1 to nearly 1:1, where ␯ is the vibrational state of the
N2 photofragment. At even higher photon energies, the ␯ϭ1
state of N2 dominates the vibrational distribution and the
population in the ␯ϭ2 and 3 states increases as well.
The peak widths and shape of the vibrational features of
the P(ET) distributions are determined by the experimental
energy resolution ͑Ϸ20 meV͒ and the rotational distribution
of the N2 and CH photofragments. For each N2 vibrational
state, the most probable value for total fragment rotational
energy, Erot , is given by
2
2
˜
˜
lar Ka level. For the C A
X A transitions, we excite
Љ
Љ
the largest feature of the R bandhead and assume that the
Kaϭ0 levels dominate this feature. D0 is the dissociation
energy, ET is the measured translational energy, and
EVR(N2) and EVR(CH) are the internal vibrational and rota-
tional energies of the N2 and CH photofragments.
D0(HNCN) for the N2 loss channel can be extracted
from these distributions so long as one can determine an
accurate value for EmT ax , the translational energy correspond-
ing to photofragments with zero internal energy. As dis-
cussed in Sec. III C, this condition is satisfied, since the dis-
tributions all drop off sharply above a consistent value of
EmT ax . For HNCN, the sharpest onset occurs at h␯ϭ3.609
eV, yielding D0ϭ2.80Ϯ0.02 eV. From our experimental
value for D0 , and from literature values for the photofrag-
ment heats of formation ͑see Table II͒, and heat capacities,47
we calculate ⌬fH0(HNCN)ϭ3.35Ϯ0.02 eV and
⌬fH298(HNCN)ϭ3.31Ϯ0.02 eV. The latter is in reasonable
agreement with ⌬fH298(HNCN)ϭ3.34Ϯ0.13 eV derived by
Clifford et al.23 but smaller error bars are associated with our
value. Similarly, from the P(ET) distribution for DNCN at
3.599 eV, D0ϭ2.835Ϯ0.030 eV and ⌬fH0(DNCN)ϭ3.33
Ϯ0.03 eV.
ErotϭEmax ␯͒ϪE
␯͒.
͑
peak
͑2͒
͑
T
Here Epeak(␯) is the translational energy of the peak maxi-
mum of a particular N2 vibrational state and EmT ax(␯) repre-
sents the maximum translational energy for that state, corre-
sponding to photofragments with zero rotational excitation
and given by
EmT ax ␯͒ϭh␯ϪD ϪEvib ␯͒,
͑3͒
͑
͑
0
N
2
where h␯ is the photon energy, D0 is the dissociation energy,
and EvNib(␯) is the energy for the ␯th vibrational state of N2.
2
The P(ET) distributions also reveal the internal energy
distributions of the nascent N2 and CH photofragments. The
resolved structure in these distributions reflects the photo-
fragment vibrational distributions. The nearly identical struc-
Table IV lists values of Erot for both HNCN and DNCN.
The vibrational structure is increasingly broadened for higher
photon energies, making a precise determination of the peak
positions less certain as reflected by the reported error bars.
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:
130.113.86.233 On: Wed, 10 Dec 2014 08:12:33

J. Chem. Phys., Vol. 114, No. 20, 22 May 2001
Isomerization and dissociation of HNCN
9007
TABLE IV. Rotational energy maxima as function of N2 vibrational quan-
tum number.
Photon energy
E
rot(␯ϭ0)
͑eV͒
Erot(␯ϭ1)
͑eV͒
Erot(␯ϭ2)
͑eV͒
͑eV͒
HNCN
DNCN
3.603
3.911
4.035
4.157
3.599
3.909
4.035
0.125Ϯ0.10
0.178Ϯ0.020 0.131Ϯ0.020 0.049Ϯ0.025
0.064Ϯ0.010
¯
¯
0.147Ϯ0.015 0.064Ϯ0.020
0.173Ϯ0.025 0.096Ϯ0.025
0.045Ϯ0.020
0.21Ϯ0.020
0.13Ϯ0.015
¯
0.174Ϯ0.020 0.075Ϯ0.020 0.031Ϯ0.020
0.210Ϯ0.030 0.081Ϯ0.030 0.056Ϯ0.030
The values of Erot exhibit a strong dependence on the N2
vibrational quantum number, with higher N2 vibrational
states yielding smaller values of Erot . For a given N2 vibra-
tional state, Erot increases with photon energy. Based on the
peak widths for each feature in the P(ET) distributions,
higher values of Erot are correlated with broader rotational
distributions.
FIG. 7. Energetics of the excited states of HNCN and schematic of the
overall dissociation reaction mechanism. The energetic positions of the
2
X
2A state is based upon this work, the term values for the A A , B 2A ,
˜
˜
˜
Љ
˜
Љ
Ј
and C 2A states are based upon the respective spectroscopic studies, Refs.
Љ
32, 24, and 26. The structure for HNCN is based upon spectroscopic results
͑Ref. 21͒, the structure and energies for TS1 are from Ref. 13, while the TS2
and INT1 are from Ref. 12.
Finally, we can place limits on the rate of dissociation
2
˜
for the excited B A state. The LIF experiments of Wu
Ј
et al.25 determined the fluorescence lifetime of the B A
2
˜
Ј
2
˜
barrier. The second observation is quite unusual and suggests
a highly specific partitioning of energy among product rota-
tional and vibrational degrees of freedom, with most of the
available internal energy channeled into N2 vibration. To our
knowledge, the only other observation of a vibrationally re-
solved P(ET) distribution for a channel with two molecular
products was in the dissociative photodetachment of OϪ4 .52
This trend is consistent with passage through a linear or cy-
clic HNCN transition state in which the N–N bond length
differs from that of diatomic N2, with a considerably smaller
difference for the C–H bond length.
→X A transition to be 20Ϯ5 ns, so the lifetime ␶ with
Љ
d
respect with respect to dissociation must at least be this long.
An upper limit on ␶ can be obtained because delayed dis-
d
sociation reduces the effective flight length between the pho-
tolysis laser and the detector ͑1 m͒. This effect transforms
what would normally be an infinitely sharp peak in the
P(ET) distribution to a feature with an exponential ‘‘tail’’
toward low ET with a width on the order of ⌬E/ET
ϭ2␶d /␶l , where ␶ ͑ϳ5 ␮s͒ is the flight time from the in-
l
teraction region to the detector. For the HNCN P(ET) distri-
bution at h␯ϭ3.603 eV, simulations of the peak shapes yield
an upper bound of ϳ250 ns for ␶d , beyond which the expo-
nential tail becomes too prominent in comparison with the
experimental peaks.
Ab initio calculations by Cui et al.,11,12 Lin and
co-workers,13,19 and Walch15 on the potential energy surfaces
of CHϩN2 have identified a number of bound intermediates
and transition states structures that may be relevant to the
dissociation mechanism of HNCN. Of particular interest are
the linear ͑lin-͒ HCNN and cyclic (c-) HCN2 isomers calcu-
lated to lie 1.2 and 1.0 eV, respectively, below the CHϩN2
product channel. The linear isomer does not have an exit
barrier with respect to dissociation to CHϩN2, so product
state distributions from this intermediate should be reason-
ably described by statistical models such as a prior
distribution,53 which, depending on the available energy, pre-
dicts between 25%–42% of the available energy is parti-
tioned into translation. This statistical distribution is in sharp
contrast to the experimental P(ET) distributions with 61%–
75% of the available energy partitioned into translation.
The reaction coordinate for dissociation of HNCN to
CHϩN2 via c-HCN2 is shown in Fig. 7. Two transition
states are involved: TS1, the transition state for isomerization
of HNCN to c-HCN2, and TS2, the transition state for dis-
sociation of c-HCN2 to CHϩN2. The TS2 structure is a rea-
sonable candidate for the transition state leading to the ex-
perimental P(ET) distributions. The tight three-center
transition state is expected to give rise to a nonstatistical
product state distribution and it is energetically accessible at
V. DISCUSSION
The formation of CHϩN2 photofragments from the
HNCN radical is an interesting and surprising result, requir-
ing both an H-atom shift to the central carbon and bending of
the NCN backbone to allow N–N bond formation, and sug-
gesting linear or cyclic isomers of HNCN as possible inter-
mediate structures. Such a complicated rearrangement pro-
cess is expected to give rise to substantial potential energy
barriers. Our goal in this section is to use the experimental
product state internal energy distributions to gain insight into
the barrier heights, transition state geometries, and electronic
states involved in the dissociation of electronically excited
HNCN to CHϩN2 products.
The most important trends in the experimental P(ET)
distributions are that ͑1͒ they all extend toward EmT ax and
show that most of the available energy goes into product
translation, and ͑2͒ they display resolved vibrational struc-
ture characteristic of the N2 photofragment. From ͑1͒ we
infer that dissociation occurs either on a repulsive surface or
through a tight transition state associated with passage over a
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:
130.113.86.233 On: Wed, 10 Dec 2014 08:12:33

9008
J. Chem. Phys., Vol. 114, No. 20, 22 May 2001
Bise, Hoops, and Neumark
TABLE V. Transition state bond angles and distance from the ab initio
calculations of Cui et al. ͑Refs. 11 and 12͒. The bond distances for the free
diatomic fragments ͑Ref. 54͒ are shown for a comparison.
dissociation to product degrees of freedom. The CH and N2
product vibrational populations are then a reflection of the
populations of the CH and N2 stretching modes at the tran-
sition state. Using the vibrational frequencies of Cui
et al.,11,12 and assuming that the vibrational modes are har-
monic, the relative population of the vibrational levels of the
transition state were calculated as a function of energy above
the barrier using the Beyer–Swinehart algorithm.58 These
calculations indicate that a statistical distribution of transi-
tion state vibrational states and a barrier height of 0.5 eV
above the dissociation asymptote reproduce the N2 vibra-
tional distribution of at 3.603 eV; this barrier height is rea-
sonable based on the energetics in Fig. 7. However, using
this same barrier height, the degree of N2 vibrational excita-
tion is underestimated at higher photon energies, suggesting
that the vibrational populations of the transition state modes
are not statistically distributed.
Overall, the experimental N2 vibrational distributions
fall between the FC and SACM model limits. The fact that
neither model works over the entire range of photon energies
is not entirely surprising. On the other hand, the SACM and
FC models reasonably reproduce the N2 vibrational distribu-
tions at the lowest and highest photon energies, respectively,
and this limited agreement provides more quantitative sup-
port for assuming TS2 to be the transition state through
which dissociation occurs.
Cyclic transition
state ͑TS2͒
Diatomic fragments
RCH
1.103 Å
1.189 Å
1.522 Å
1.883 Å
81.6°
1.1199 Å
1.0976 Å
RNN
RCN1
RCN2
␪
CH
␪
53.8°
CN2
Dih H–C–N2–N1
Ϫ158.4°
the photon energies used in this study. The structural param-
eters for TS2 calculated by Cui et al.11,12 at the UCCSD͑T͒/
6-311G͑d,p͒ level of theory are listed in Table V, along with
those of diatomic N2 and CH; the change in N–N bond
length upon dissociation is considerably larger than the
change in C–H bond length, providing further motivation for
considering TS2 as the relevant transition state for our ex-
periment.
In order to be more quantitative, we compare below our
experimental product state distributions with those expected
from simple dissociation models involving dissociation via
the TS2 transition state. The product vibrational and rota-
tional distributions are treated separately.
A. Vibrational distributions
B. Rotational distributions—modified impulsive
model
In this section, limiting models for the product vibra-
tional distributions are considered assuming TS2 to be the
transition state for dissociation. In the limit of rapid disso-
ciation over a barrier the steep repulsive exit valley does not
allow for coupling between vibrational and translational
modes. A simple model for describing the resulting photo-
fragment vibrational distribution is provided by a Franck–
Condon ͑FC͒ mapping of the transition state equilibrium ge-
ometry onto the photofragment equilibrium bond distances.55
As is seen in Table V, the CH bond length changes only
0.017 Å from TS2 to the free diatomic. FC mapping of the
transition state produces Ͼ99% of the vibrational population
in ␯ϭ0. The N2 bond distance shows a more substantial
change, decreasing by ϳ0.09 Å from TS2, leading to a N2
vibrational distribution that peaks at ␯ϭ1 with nearly equal
populations in ␯ϭ1 and ␯ϭ2. Table III shows that a FC
mapping of the transition state, which is independent of pho-
ton energy, overestimates the extent of N2 vibrational exci-
tation compared to the P(ET) distribution from excitation at
3.603 eV, although it agrees well with the results at 4.157
eV.
The results in Table IV show that at each photon energy,
Erot decreases with increasing N2 vibration. This trend sug-
gests that the rotational distributions should be analyzed us-
ing the modified impulsive model,59–61 which assumes that
dissociation is mediated by an instantaneous repulsive force
between the departing atoms and that the vibrations of the
recoiling fragments are infinitely stiff. The impulse does not
generate fragment vibrational excitation. The fragment vibra-
tional distribution is formed prior to the impulse, and hence
this vibrational energy, Evib , is not available to translation or
rotation. The impulsive fragment rotational energy, Eirmotp , de-
scribed by Eq. ͑4͒, is dependent upon the parent molecular
geometry and the energy available for translation,
imp
E
ϭ␣• EЈ ͒ϩ␤h␯
,
͑4͒
͑
rot
avail
where ␣ describes the geometric factors as described by
Buttenhoff et al.,62 EaЈvail is equal to the total energy available
to the products minus the fragment vibrational energy, and
␤
is ideally equal to zero.
h␯
The first step in assessing the suitability of this model is
The observed evolution of the P(ET) distributions to-
ward higher N2 vibration with increasing photon energy sug-
gests that higher vibrational levels of the transition state be-
come populated as h␯ increases. An alternative description
of the product vibrational distribution that accounts for this
effect is the statistically adiabatic channel model
͑SACM͒.56–58 In this model it is assumed that the vibrational
levels of the transition state are statistically populated and
that the vibrational quantum numbers are preserved through
to determine how well the experimental data in Table IV are
described by Eq. ͑4͒. The parameter ␣ was found at each
photon energy and the results are shown in Fig. 8 for HNCN
and DNCN. This plot shows that the entire dataset for each
species has approximately the same value of ␣, with slightly
higher values, in general, for DNCN than HNCN. However,
plots of Erot vs Eavail have small, nonzero y-intercept values
␤
that vary between Ϫ0.04 to Ϫ0.13 eV for HNCN and
h␯
Ϫ0.10 to Ϫ0.14 eV for DNCN, in contrast to the predictions
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:
130.113.86.233 On: Wed, 10 Dec 2014 08:12:33

J. Chem. Phys., Vol. 114, No. 20, 22 May 2001
Isomerization and dissociation of HNCN
9009
FIG. 8. Experimental values of ␣ for the dependence of Erot with Eavail for
HNCN ͑᭺͒ and DNCN ͑᭝͒ at various photon energies ͑see the text͒. Cal-
culated values from the modified impulsive model described in the text are
shown with dashed horizontal lines. The error bars are significantly larger
for data points at 3.6 and 4.0 eV since only two data points were used to
determine the slope.
FIG. 10. A comparison of the experimental P(ET) distribution for excita-
tion at 3.6027 eV ͑᭺͒ with the impulsive model calculation. The dashed–
dot line is the result of the impulsive model including the zero-point vibra-
tional motion of the transition state. The solid line includes the zero-point
motion and an estimated parent rotational distribution of 100 K.
of the model. Nonetheless the relatively small variations in
slope ͑i.e., ␣͒ suggest that the model be pursued further.
The next step is to compare the experimental values of ␣
to predictions from the modified impulsive model, again as-
suming TS2 as the transition state for dissociation. For
simple bond cleavage, the impulsive force is naturally ap-
plied along the breaking bond, but for three-center-type dis-
sociations, such as c-HCN2→CHϩN2, the impulse need not
occur along an individual bond.62 The normal modes of TS2
are shown in Fig. 9.63 An approximate direction of the im-
pulse can be obtained from the motion along the reaction
tofragment, e.g., a value of EЈavailϭ0.6 eV gives rise to
rotational quantum numbers of 19 and 3 for jN and jCH
,
2
respectively. Values of ␣ from the model calculation for
HNCN and DNCN are shown as horizontal lines in Fig. 8.
The larger value for the DNCN system is due to the larger
impact parameter associated with the CD fragment. There is
reasonable agreement with the experimental values for both
species, supporting the validity of the model and the assump-
tion of TS2 as the transition state.
The modified impulsive model provides only a single
rotational energy, not a rotational distribution. Such a distri-
bution can be obtained by including the effect of parent vi-
brational and rotational motion.60 We first consider the con-
tribution from zero-point vibrational motion. Using the
eigenvectors and frequencies for the transition state normal
modes63 ͑Fig. 9͒, and following the work of Buttenhoff
et al.,62 we have determined the contribution of the zero-
point motion of each transition state vibration toward the
fragment rotational distributions. Vibrational motion in the
CNN plane will have the largest effect upon the rotational
distribution since the momentum will add or subtract from
the impulsive momentum. Asymmetric stretching of the
C–N bonds, ␯4 , leads to the largest spread in the N2 rota-
tional distribution with a FWHM of four rotational quanta
while the in-plane H-atom bending mode, ␯3 , leads to a
FWHM of 2 quanta for the CH fragment. Converting the
rotational distribution into an energy distribution and then
convoluting with our experimental energy resolution yields
the dashed line in Fig. 10 for the P(ET) distribution at h␯
ϭ3.603 eV. It appears that vibrational motion does not com-
pletely account for the total width of the experimental trans-
lational energy distribution. However, following Levene and
Valentini,10 parent rotational excitation is also expected to
contribute to the fragment rotational distribution resulting in
further broadening. A temperature of 100 K for the rotational
distribution of the transition state provides a reasonable fit to
the experimental data, shown with a solid line in Fig. 10.
coordinate, the ␯ eigenvector, which is primarily associated
6
with stretching of the long C–N bond (C–N2). As pointed
out by Buttenhoff et al.,62 the impulsive force occurs in the
exit valley past the transition state geometry since at the
transition state itself, there is no force on the photofragments.
Here, we have made the further approximation that the ge-
ometry at which the impulse is applied is that of the transi-
tion state. The impulsive force primarily excites the N2 pho-
FIG. 9. Eigenvectors and eigenvalues of the normal modes for TS2 from the
ab initio calculations of Cui et al. ͑Ref. 58͒. Only the motion in the
C–N1–N2 plane is shown here; ␯ and ␯ possess ‘‘out-of-plane’’ H-atom
5
6
motion.
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:
130.113.86.233 On: Wed, 10 Dec 2014 08:12:33

9010
J. Chem. Phys., Vol. 114, No. 20, 22 May 2001
Bise, Hoops, and Neumark
The same procedure could not account for the width of
rather by subsequent chemistry of the NCN radical. It would
clearly be of great interest to perform experiments on the
CHϩN2 reaction under single-collision conditions to directly
determine the identity of the products of the biomolecular
reaction.
the features in the P(ET) distributions at higher photon en-
ergies. This discrepancy may reflect the contribution of vi-
brationally excited levels of the transition state, since the
larger momentum associated with higher vibrational levels
should lead to a broader fragment rotational distribution. For
example, the ␯4ϭ1 transition state level results in a rota-
tional distribution with a FWHM of seven rotational quanta
for the N2 fragment. This corresponds to a width of 170 meV
for the N2 (␯ϭ0) feature of the P(ET) distribution at 3.911
eV, accounting for most of the observed width of 210 meV.
VI. CONCLUSIONS
The photodissociation dynamics and spectroscopy of the
HNCN radical have been investigated. Photofragment yield
2
˜
spectra show that the B A excited state predissociates. In
Ј
addition, predissociation is seen from the previously ob-
served absorption bands 31 500–34 500 cmϪ1. These bands
are conclusively attributed to HNCN and are assigned as the
C. Dissociation mechanism
The preceding discussion strongly implies that the
CHϩN2 products are formed by passage through the cyclic
TS2 transition state. This raises the question of how elec-
tronically excited HNCN radicals access this transition state.
Several observations are relevant here. The PFY spectra for
2
2
˜
˜
C A
X A band. At all photon energies, CHϩN2 is
Љ
Љ
found to be the primary dissociation channel. The product
state distributions are vibrationally resolved, showing vibra-
tional progressions in the N2 fragment. The distributions are
highly nonstatistical and show an inverse relationship be-
tween rotational and vibrational excitation.
2
2
˜
˜
excitation to the B A and C A states are structured, in-
Ј
Љ
dicating predissociation rather than direct dissociation on re-
˜
pulsive surfaces. The B state, in particular, is quite long
The product vibrational and rotational distributions can
be reasonably well fit by applying simple models to the cy-
clic HCN2 transition state geometry calculated by Cui and
Morokuma,11 suggesting that the dissociation pathway to
CHϩN2 involves a c-HCN2 intermediate that dissociates via
a tight-three-center transition state. Based upon the variation
of the product state distributions with photon energy, a dis-
sociation mechanism is proposed in which the excited states
undergo internal conversion to the ground state, followed by
isomerization to c-HCN2, which then dissociates to CHϩN2.
Our results are generally consistent with ab initio barrier
heights for the CHϩN2 reaction and imply that the HNCN
radical is an important intermediate in the formation of
prompt NO in a combustion environment.
lived, with a lifetime between 20 and 250 ns ͑Sec. IV B͒.
˜
˜
The B and C excited states both dissociate to ground state
CHϩN2 products. Furthermore, the extent of N2 vibrational
excitation and width of the rotational distributions appear to
increase smoothly with excitation energy and do not display
any obvious dependence on either the electronic or vibra-
tional character of the optically prepared excited state.
This set of observations suggests that electronically ex-
cited HNCN undergoes internal conversion ͑IC͒, forming
2
˜
highly vibrationally excited HNCN in its ground X A state.
Љ
Once this occurs, dissociation to CHϩN2 can proceed along
the reaction coordinate shown in Fig. 7, in which isomeriza-
tion via TS1 to c-HCN2 is followed by dissociation via TS2
to CHϩN2 products:
IC
ACKNOWLEDGMENTS
˜
˜
˜
͑
HNCN B or C͒ —— HNCN X͒
͑
→
This research is supported by the Director, Office of En-
ergy and Research, Office of Basic Energy Sciences, Chemi-
cal Sciences Division, of the U.S. Department of Energy
under Contract No. DE-AC03-76F00098. R.T.B. would like
to thank Professor C. W. Mathews for stimulating discus-
sions and for generously providing absorption data prior to
publication, Dr. Qiang Cui and Professor Keiji Morokuma
for providing the eigenvectors of the cyclic transition state
normal modes, and L. V. Moskoleva and Professor Lin for
providing a copy of their manuscript prior to publication.
isomerization
dissociation
—— c-HNCN —— CHϩN .
→
→
2
This mechanism means that our P(ET) distributions are
largely a result of ground state dynamics and are therefore
relevant to the combustion chemistry of the CHϩN2 reac-
tion. The barrier heights for TS1 and TS2 calculated by Cui
et al.11,12 and Moskaleva et al.19 must be approximately cor-
˜
rect, because if they were much higher the B state would not
be able to dissociate to CHϩN2 products ͑see Fig. 7͒. As a
consequence, our results imply that at high temperature, the
CHϩN2 reaction can access the HNCN intermediate. We do
not observe the spin-forbidden N(4S)ϩHCN channel, indi-
cating that intersystem crossing subsequent to IC does not
occur to any significant extent. Our results therefore indi-
rectly support the recent proposal by Moskaleva and Lin13–19
that the CHϩN2 reaction does not yield N(4S)ϩHCN, but
instead results in higher-energy, spin-allowed H(2S)
ϩNCN(X ͚g ) products through dissociation of the HNCN
radical. As discussed in the Introduction, this proposal im-
plies that in combustion, prompt NO is not formed by oxi-
dation of N atoms produced in the CHϩN2 reaction but
1 J. A. Miller and C. T. Bowman, Prog. Energy Combust. Sci. 15, 287
͑1989͒.
2 J. A. Miller and S. P. Walch, Int. J. Chem. Kinet. 29, 253 ͑1997͒.
3 M. R. Berman and M. C. Lin, J. Phys. Chem. 87, 3933 ͑1983͒.
4 L. J. Medhurst, N. L. Garland, and H. H. Nelson, J. Phys. Chem. 97,
12 275 ͑1993͒.
5 A. J. Dean, R. K. Hanson, and C. T. Bowman, in Proceedings of the 23rd
Symposium (International) on Combustion ͑The Combustion Institute,
Pittsburgh, PA, 1990͒, p. 259.
Ϫ
3
˜
6 C. P. Fenimore, in Proceedings of the 13th Symposium (International) on
Combustion ͑The Combustion Institute, Pittsburgh, PA, 1971͒, p. 373.
7 J. A. Miller, M. C. Branch, W. J. McLean, D. W. Chandler, M. D.
Smooke, and R. J. Kee, in Proceedings of the 20th Symposium (Interna-
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:
130.113.86.233 On: Wed, 10 Dec 2014 08:12:33

J. Chem. Phys., Vol. 114, No. 20, 22 May 2001
Isomerization and dissociation of HNCN
9011
tional) on Combustion ͑The Combustion Institute, Pittsburgh, PA, 1984͒,
p. 373.
36 D. L. Osborn, H. Choi, D. H. Mordaunt, R. T. Bise, D. M. Neumark, and
C. M. Rohlfing, J. Chem. Phys. 106, 3049 ͑1997͒.
37 D. L. Osborn, D. J. Leahy, D. R. Cyr, and D. M. Neumark, J. Chem. Phys.
104, 5026 ͑1996͒.
8 D. Lindackers, M. Burmeister, and P. Roth, 23rd International Symposium
on Combustion, 1990, Vol. 23, p. 251 ͑1990͒.
9 M. R. Manaa and D. R. Yarkony, J. Chem. Phys. 95, 1808 ͑1991͒.
10 M. R. Manaa and D. R. Yarkony, Chem. Phys. Lett. 188, 352 ͑1992͒.
11 Q. Cui and K. Morokuma, Theor. Chem. Acc. 102, 127 ͑1999͒.
12 Q. Cui, K. Morokuma, J. M. Bowman, and S. Klippenstein, J. Chem.
Phys. 110, 9469 ͑1999͒.
38 T. R. Taylor, R. T. Bise, K. R. Asmis, and D. M. Neumark, Chem. Phys.
Lett. 301, 413 ͑1999͒.
39 J. M. B. Bakker, J. Phys. E 6, 785 ͑1973͒.
40 J. M. B. Bakker, J. Phys. E 7, 364 ͑1974͒.
41 D. P. de Bruijn and J. Los, Rev. Sci. Instrum. 53, 1020 ͑1982͒.
42 G. Herzberg, Molecular Spectra and Molecular Structure III. Electronic
Spectra and Electronic Structure of Polyatomic Molecules ͑Van Nostrand
Reinhold, New York, 1966͒.
13 L. V. Moskaleva and M. C. Lin, ‘‘Colloquium on NOx , SOx and pollutant
emission kinetics,’’ 2000.
14 J. M. L. Martin and P. R. Taylor, Chem. Phys. Lett. 209, 143 ͑1993͒.
15 S. P. Walch, Chem. Phys. Lett. 208, 214 ͑1993͒.
16 T. Seideman and S. P. Walch, J. Chem. Phys. 101, 3656 ͑1994͒.
17 S. P. Walch, J. Chem. Phys. 103, 4930 ͑1995͒.
43 P. S. H. Bolman and J. M. Brown, Chem. Phys. Lett. 21, 213 ͑1973͒.
44 R. N. Dixon, Can. J. Phys. 38, 10 ͑1960͒.
45 R. N. Dixon, M. J. Trenouth, and C. M. Western, Mol. Phys. 60, 779
͑1987͒.
18 T. Seideman, J. Chem. Phys. 101, 3662 ͑1994͒.
19 L. V. Moskaleva, W. S. Xia, and M. C. Lin, Chem. Phys. Lett. 331, 269
͑2000͒.
46 S. A. Wright and P. J. Dagdigian, J. Chem. Phys. 104, 8279 ͑1996͒.
47 M. W. Chase, Jr., J. Chem. Phys. 9, 1 ͑1998͒.
20 R. T. Bise, H. Choi, and D. M. Neumark, J. Chem. Phys. 111, 4923
͑1999͒.
48 M. Danielsson, P. Erman, A. Hishikawa, M. Larsson, R.-K. E. Sumame,
21 R. T. Bise, A. A. Hoops, and D. M. Neumark, J. Chem. Phys. 113, 1
͑2000͒.
¨
and G. Sundstrom, J. Chem. Phys. 98, 9405 ͑1993͒.
49 Y. Huang, S. A. Barts, and J. P. Halpern, J. Phys. Chem. 96, 425 ͑1992͒.
50 E. P. Clifford, P. G. Wenthold, W. C. Lineberger, G. A. Petersson, K. M.
Broadus, S. R. Kass, S. Kato, C. H. DePuy, V. M. Bierbaum, and G. B.
Ellison, J. Phys. Chem. A 102, 7100 ͑1998͒.
22 R. T. Bise, A. A. Hoops, H. Choi, and D. M. Neumark, in Imaging in
Chemical Dynamics, edited by A. G. Suits and R. E. Continetti ͑American
Chemical Society, Washington, DC, 2000͒, p. 296.
23 E. P. Clifford, P. G. Wenthold, W. C. Lineberger, G. A. Petersson, and G.
B. Ellison, J. Phys. Chem. A 101, 4338 ͑1997͒.
51 A. Hansel, C. Scheiring, M. Glantschniq, and W. Lindinger, J. Chem.
Phys. 109, 1748 ͑1998͒.
24 G. Herzberg and P. A. Warsop, Can. J. Phys. 41, 286 ͑1963͒.
25 W. Wu, G. Hall, and T. J. Sears, J. Chem. Soc., Faraday Trans. 89, 615
͑1993͒.
52 C. R. Sherwood, M. C. Garner, K. A. Hanold, K. M. Strong, and R. E.
Continetti, J. Chem. Phys. 102, 6949 ͑1995͒.
53 R. D. Levine, in Theory of Chemical Reaction Dynamics, edited by M.
Baer ͑CRC, Boca Raton, FL, 1985͒, Vol. IV, p. 1.
26 G. Herzberg and D. N. Travis, Can. J. Phys. 42, 1658 ͑1964͒.
27 N. Basco and K. K. Yee, Chem. Commun. ͑Cambridge͒ 3, 150 ͑1968͒.
28 H. W. Kroto, T. F. Morgan, and H. H. Sheena, Trans. Faraday Soc. 66,
2237 ͑1970͒.
54 K. P. Huber and G. Herzberg, Constants of Diatomic Molecules ͑Van
Nostrand Reinhold, New York, 1979͒.
55 R. Schinke, Photodissociation Dynamics ͑Cambridge University Press,
Cambridge, 1993͒.
29 C. W. Mathews, B. S. Pappas, and D. A. Ramsay, in NRC Spectroscopy
Conference in Memory of Gerhard Herzberg, Ottawa, Canada, 1999.
30 S. Yamamoto and S. Saito, J. Chem. Phys. 101, 10 350 ͑1994͒.
31 G. Maier, J. Eckwert, A. Bothur, H. P. Reisenauer, and C. Schmidt, Lie-
bigs Ann. 7, 1041 ͑1996͒.
56 M. Quack and J. Troe, Ber. Bunsenges. Phys. Chem. 58, 240 ͑1974͒.
57 M. Quack and J. Troe, Ber. Bunsenges. Phys. Chem. 79, 469 ͑1975͒.
58 T. Baer and W. L. Hase, Unimolecular Reaction Dynamics ͑Oxford Uni-
versity Press, New York, 1996͒.
32 Z. Liu, R. T. Bise, and D. M. Neumark ͑unpublished͒.
33 F.-M. Tao, W. Klemperer, M. C. McCarthy, C. A. Gottlieb, and P. Thad-
deus, J. Chem. Phys. 100, 3691 ͑1994͒.
59 G. E. Busch and K. R. Wilson, J. Chem. Phys. 56, 3626 ͑1972͒.
60 H. B. Levene and J. J. Valentini, J. Chem. Phys. 87, 2594 ͑1987͒.
61 A. F. Tuck, J. Chem. Soc., Faraday Trans. 73, 689 ͑1977͒.
62 T. J. Buttenhoff, K. L. Carleton, and C. B. Moore, J. Chem. Phys. 92, 378
͑1990͒.
34 D. J. Leahy, D. L. Osborn, D. R. Cyr, and D. M. Neumark, J. Chem. Phys.
103, 2495 ͑1995͒.
35 R. E. Continetti, D. R. Cyr, R. B. Metz, and D. M. Neumark, Chem. Phys.
Lett. 182, 406 ͑1991͒.
63 Q. Cui and K. Morokuma ͑private communication͒.
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:
130.113.86.233 On: Wed, 10 Dec 2014 08:12:33