Available online at www.sciencedirect.

com

Chemical Physics 347 (2008) 288–299

www.elsevier.com/locate/chemphys

Further aspects of the roaming mechanism in formaldehyde dissociation

S.A. Lahankar a, V. Goncharov a, F. Suits b, J.D. Farnum c,
J.M. Bowman c, Arthur G. Suits a,*
a
Department of Chemistry, Wayne State University, Detroit, MI 48202, United States
b
Computational Biology Center, IBM T. J. Watson Research Center, Yorktown Heights, NY 10598, United States
c
Department of Chemistry and Cherry L Emerson Center for Scientific Computation, Emory University, Atlanta, GA 30322, United States

Received 26 September 2007; accepted 6 November 2007

Available online 22 November 2007

Abstract

Recently we reported a novel ‘‘roaming” dissociation pathway of formaldehyde in which one of the H atoms strays far from the min-
imum energy reaction path, explores a broad region of the potential energy surface, then abstracts the remaining H atom to form molec-
ular products, without going near the configuration of the conventional transition state saddle point. The detailed dynamics of the
abstraction mechanism and its energy dependence have already been reported. Here, with a combination of experimental and theoretical
results, we examine the roaming behavior at the energetic extremes. We show evidence of roaming below the threshold of the radical
dissociation channel and consider the implications of this and the possible existence of a transition state for the roaming mechanism.
We also show the occurrence of roaming up to 3200 cm1 above the threshold of the triplet dissociation channel. In addition, we pres-
ent results affording deeper insight into the dynamics of the roaming mechanism: we show evidence of roaming leading to CO in v = 1,
and examine the issue of nuclear spin conservation during dissociation via the roaming mechanism.
Ó 2007 Elsevier B.V. All rights reserved.

Keywords: Photodissociation; Reaction dynamics; Excited states; Roaming atoms; Photochemistry

1. Introduction H2 COðT1 Þ ! H þ HCO ðChannel 4Þ

H2 COðT1 Þ ! H þ H þ CO ðChannel 5Þ
Formaldehyde has been a prototype molecule in the
H2 COðS1 Þ ! H þ H þ CO ðChannel 6Þ
study of spectroscopy and reaction mechanisms for dec-
ades [1–16]. It was the first polyatomic molecule on which In the 1980s and early 1990s, the Moore group at Berkeley
a detailed spectroscopic investigation was carried out [1,2]. investigated the dissociation of formaldehyde via the molec-
The relevant energy levels are given in Fig. 1. Its dissocia- ular transition state (TS) to form CO and H2 (channel 1) in
tion processes following excitation in the near ultraviolet, detail [8,10,17–26]. The dissociation of formaldehyde via
which take place after internal conversion to the ground channel 1 yields rotationally hot CO (jCO = 21–70) and vib-
state or intersystem crossing to the triplet, can be summa- rationally ‘‘cold” H2(mH2 = 0–5) [13,27]. The observation of
rized as the following reactions [16]: rotationally hot CO was assigned [20] to impulsive dissoci-
ation via the TS, which has a skewed planar geometry with
H2 COðS0 Þ ! H2 þ CO ðChannel 1Þ
both H atoms on the same side of the CO [28–32]. The bar-
H2 COðS0 Þ ! H þ HCO ðChannel 2Þ rier for this dissociation is 27,720 cm1 [10]. Recently
H2 COðS0 Þ ! ½H    HCO ! H2 þ CO ðChannel 3Þ Chambreau et al. reported a state-correlated investigation
of the dissociation of formaldehyde via channel 1 [27]. A
strong correlation between the rotational states of CO
*
Corresponding author. Tel.: +1 631 632 1702; fax: +1 631 632 7960. and the vibrational states of H2 was observed up to
E-mail address: asuits@chem.wayne.edu (A.G. Suits). 4100 cm1 above the barrier for channel 1. This strong

0301-0104/$ - see front matter Ó 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.chemphys.2007.11.007
 S.A. Lahankar et al. / Chemical Physics 347 (2008) 288–299 289

In 2004, our group, in collaboration with the Bowman

group at Emory, identified a new dissociation pathway of
formaldehyde and termed it the roaming mechanism [13].
This reaction, anticipated in earlier studies [10], from the
Moore group, starts as channel 2, although it lacks suffi-
cient energy in the reaction coordinate to dissociate. One
of the H atoms (to be denoted as HR) roams around the
HCO until it abstracts the remaining H (HL) atom from
HCO to form vibrationally hot H2 (mH2 = 5–8) and rota-
tionally cold CO (jCO = 0–32) [13,14,38]. The dynamics of
this H-abstraction have been examined in detail. The result-
ing high vibrational excitation in H2 is due to the exoergic
reaction occurring at long H–H distances. The HL tends
to leave HCO along its equilibrium bond angle irrespective
of the position of HR relative to the HCO. This leads to sig-
Fig. 1. Formaldehyde dissociation thresholds along with energy levels
nificant rotational excitation in product H2 from roaming
accessed in this investigation.
[14]. The roaming mechanism is a large amplitude motion
that occurs in the H–HCO region of configuration space,
correlation is mainly dynamical in origin, a consequence of with of course a minimum flux dividing surface and there-
the highly repulsive exit channel and the skewed geometry fore a (variational) transition state. Thus, H2 + CO can be
of the TS [27]. It was also found that the correlation be- formed from two quite different pathways through two
tween the rotational states of H2 and CO was very weak transition states; these two pathways are denoted as chan-
as noted earlier by Moore and coworkers. nels 1 (conventional) and channel 3 (roaming).
Moore’s group also investigated the conservation of The roaming mechanism (channel 3) occurs on the sin-
nuclear spin in formaldehyde during the molecular dissoci- glet ground state and competes with channel 1 and channel
ation via the 2141 transition [18]. Formaldehyde is a planar 2. This was examined in detail experimentally and theoret-
molecule with C2v point group. Two nuclear spin modifica- ically for excitation via the 2143 transition [14]. The overall
tions are possible for formaldehyde: ortho-formaldehyde branching between channel 3 and channel 1 was found to
and para-formaldehyde. ortho-Formaldehyde has odd Ka be 0.18:0.82. This was then used to estimate the energy
values and para-formaldehyde has even Ka values. Moore dependence of the branching between channels 1 and 3
et al. reported that nuclear spin is conserved between obtained from excitation spectra [15]. We found that as
formaldehyde and H2 during the dissociation. That is, the the available energy increases, channel 3 grows significantly
dissociation of ortho-formaldehyde leads to formation of relative to channel 1 [15]. These estimated branching frac-
ortho-H2 and the dissociation of para-formaldehyde leads tions were combined with the theoretical estimates of
to formation of para-H2. branching of molecular products relative to radical prod-
Formaldehyde also undergoes simple bond fission to ucts to yield overall multichannel branching. The branch-
form radical products H and HCO (Channel 2). This reac- ing results showed that as the available energy is
tion is barrierless on the ground singlet surface. Kable and increased from 30,300 to 32,000 cm1, the contribution of
co-workers [11,33–37] and Wittig and co-workers [12] have channel 1 drops from 0.8 to 0.15, the contribution of chan-
studied the dynamics of this process. The threshold for nel 2 increases from 0.0 to 0.75, and the contribution from
channel 2 is 30327.6 cm1 [37]. Dissociation via channel 2 channel 3 decreases from 0.18 to 0.10 [15].
yields high N HCO. The product HCO distribution from In the present paper, we bracket the threshold for chan-
channel 2 is statistical and successfully modeled by phase nel 3, thus determine the effective energy width of the
space theory (PST). Radical dissociation of formaldehyde roaming event and the energy of a possible saddle point
on the triplet surface (Channel 4) has a threshold of ‘‘region” for the roaming pathway recently suggested by
32 000 cm1, implying an exit barrier of 1700 cm1 Harding and coworkers [39]. This is considered in light of
[33]. However there is possible tunneling through the bar- quasiclassical trajectory (QCT) calculations on the fitted
rier of channel 4 following excitation to the 2243 transition surface. We also report evidence of roaming leading to vib-
at 31,555 cm1 [34]. Dissociation of formaldehyde via rationally excited product CO, explore roaming at very
channel 4 leads to low N HCO [33]. In this case, the prod- high excitation energies, and examine the issue of nuclear
uct state distribution (PSD) is not statistical and cannot be spin conservation during roaming dissociation.
described by PST. It was modeled as an impulsive dissoci-
ation similar to the approach used by Moore and co-work- 2. Experiment
ers for dissociation via channel 1. Recently Kable and
Bowman and their co-workers have investigated the Formaldehyde for the experiment was prepared in situ
dynamics of dissociation of formaldehyde via channel 2 with scheme used by Kable and co-workers [37], with mod-
and channel 4 in detail [33,34]. ification to suit our experimental setup. A cylindrical tube
290 S.A. Lahankar et al. / Chemical Physics 347 (2008) 288–299

with several layers of paraformaldehyde and magnesium normal modes and the excess energy was put into the imag-
sulfate separated by glass wool was connected to a piezo- inary normal mode in the form of kinetic energy. Approx-
electric pulsed valve assembly and the entire system was imate quantum states are assigned to the products of each
heated to 60 °C using electrical heating tapes. The formal- trajectory using histogram binning of the classical
dehyde was seeded in two atm Ar and supersonically quantities.
expanded into the source region. The beam was skimmed
before entering the main chamber. Typical rotational tem- 4. Results
peratures under these conditions are 10 K.
Formaldehyde photolysis was carried in the range Fig. 2 shows a high resolution PHOFEX scan of form-
30,000–35,400 cm1. The dissociation laser beam was pro- aldehyde when the probe laser was set at jCO 45, 15, and
duced by doubling the output of a tunable dye laser 10. We showed previously that when CO is probed at jCO
pumped by the second harmonic of a nanosecond neodym- 45, only dissociation via channel 1 is observed, whereas
ium: yttrium aluminum garnet (Nd:YAG) laser. The pho- when CO is probed at jCO 15 or 10, only dissociation via
tolysis energy was 1.5–2.0 mJ/pulse. The product CO channel 3 is observed [15]. Hence, PHOFEX at jCO 45 is
was probed using (2+1) resonance enhanced multiphoton used as a proxy for dissociation via channel 1 and PHO-
ionization (REMPI) via the B-state around 230 nm. The FEX at jCO 15, 10 are used as proxies for dissociation via
probe laser beam was generated by sum frequency mixing channel 3. PHOFEX at jCO 45 shows three vibrational
of the third harmonic of an injection-seeded Nd:YAG laser bands of S1 in this region: 45, 2161, 2143. However, PHO-
with the fundamental of a dye laser (654 nm) pumped by FEX scans at jCO 15 and 10 show only the two higher
the second harmonic of the same Nd:YAG laser. The energy vibrational bands of S1, 2161 and 2143; the 45 band
probe energy used was 2.5–3.0 mJ/pulse. is missing. This implies that the threshold for the roaming
Photolysis of formaldehyde followed by ionization of pathway is between the 45 band (30,123 cm1) and the 2161
CO occurs between the repeller and first lens in the ion transition (30,241 cm1). The vertical dotted line shows the
optics assembly. The ionized CO is then accelerated out experimental threshold of channel 2.
of the interaction region and detected using our high-reso- To further verify the PHOFEX scans, images were
lution DC slice imaging technique with a detector gate acquired after excitation of formaldehyde to rR1(1) branch
width of 60 ns. Details of this setup have been reported pre- of 45 and pP1(1) branch of 2161 bands, with CO probed at
viously [40]. Detected ions were imaged by a charge-cou- jCO 45 and 14. Fig. 3 shows these images and the corre-
pled device (CCD) camera and images were acquired and sponding translational energy distributions. In Fig. 3a
integrated with our IMACQ Megapixel imaging software and b, the vibrational distribution peaks at m = 2. This
[41]. The raw images were analyzed to yield product veloc- internal energy distribution of H2 is similar to the internal
ity distributions. Explicit treatment of finite slicing was also distribution obtained previously via the 2143 transition
applied to account for the difference in slice width for slow probed at jCO 45, and it clearly represents dissociation
versus fast photofragments [42]. The PHOFEX scans were via channel 1. In Fig. 3c the internal energy distribution
acquired using a photomultiplier tube viewing the detector of H2 peaks at m = 7, jH2 = 7 and m = 6, jH2 = 9. This is
in conjunction with a digital oscilloscope, with the MCP
gate expanded to 320 ns so that the full CO+ ion signal
was acquired. During scans, the laser wavelength was con-
tinuously monitored using a wavemeter. The photolysis
and probe laser power was stable during all the PHOFEX
scans and no power correction was required.

3. Theory

Investigations of the roaming region used the previously

reported potential energy surface (PES) for formaldehyde
[43]. The surface combines 15,668 CCSD(T) symmetry-
unique energies with about 46,000 MR-CI symmetry-
unique energies all calculated with the aug-cc-pVTZ basis
set. The energies are fit with a set of Morse variables to
obtain an analytic approximation to the PES of the system.
Quasi-classical trajectories have been simulated and
reported on this system and are used in the detailed inves-
tigation of the present work [13,38,44]. The trajectories
started at an initial geometry where one of the CH bonds
has been stretched to about twice the equilibrium length. Fig. 2. PHOFEX spectra obtained with CO probe was set at indicated
Zero-point energy was then randomized among the real rotational level.
 S.A. Lahankar et al. / Chemical Physics 347 (2008) 288–299 291

Fig. 3. Images (left) and translational energy distributions (right) acquired at: (a) 45 jCO 45; (b) 2161 jCO 45; (c) 2161 jCO 14.

similar to the internal energy distribution obtained for low goes up to m = 10, jH2 = 7. For the 2441 level (Fig. 4c) the
jCO via the 2143 band, and can be explained by the roaming internal energy distribution peaks at H2m = 7, 8 and goes
model. No signal was observed for the 45 band when CO is up to m = 9, jH2 = 7.
probed at jCO 15 or 10. Images and the corresponding translational energy dis-
A number of images and associated translational energy tributions for low rotational levels of CO (v = 1) acquired
distributions obtained for low rotational levels of CO after excitation of formaldehyde via the 2441 and 2143
(v = 0) following photolysis at 35,217 cm1 via the 2641 bands are shown in Fig. 5. The internal energy distribution
and 2441 bands are shown in Fig. 4. Fig. 4a and b show that of H2 peaks at m = 7, 8, jH2 = 7 for the 2441 band whereas it
the internal energy distribution of H2 peaks at m = 8, 9 and peaks at m = 7, jH2 = 3 and m = 6, jH2 = 7 for 2143 band.
292 S.A. Lahankar et al. / Chemical Physics 347 (2008) 288–299

Fig. 4. Images (left) and translational energy distributions (right) acquired at: (a) 35,217 cm1 jCO 12; (b) 2641 jCO 18; (c) 2441 jCO 15.

The images and associated translational energy distribu- band. In the translational energy distributions, the dark
tions in Fig. 6 were acquired by excitation of formaldehyde curve is the result of para-formaldehyde dissociation
to the 2241 level and individually selecting (a) ortho and (b) whereas the light curve is from ortho-formaldehyde dissoci-
para-formaldehyde. The ortho-formaldehyde is selected by ation. The combs show the rovibrational assignment for
excitation to the rR1(1) branch of 2241 and para-formalde- dissociation to jH2 = 6, 8 for para H2 and jH2 = 7,9 for
hyde is selected by excitation to rR0(0) branch of the 2241 ortho H2.
 S.A. Lahankar et al. / Chemical Physics 347 (2008) 288–299 293

Fig. 5. Images (left) and translational energy distributions (right) acquired at: (a) 2441m = 1, jCO 14; (b) 2143 m = 1, jCO 15.

5. Discussion high H2 vibrational excitation [14], and we showed that

the abstraction reaction favors transfer of the HCO H
These new results probe several distinct aspects of the atom along the HCO equilibrium direction regardless of
roaming mechanism in formaldehyde dissociation. In pre- the approach of the roaming atom. Furthermore, we found
vious work we have shown this pathway to be the result no strong constraint for roaming in the HCO plane. The
of frustrated dissociation via the radical channel that leads PHOFEX spectra (Fig. 2) and associated images and trans-
to intramolecular H abstraction and distinct product state lational energy distributions (Fig. 3) obtained below the
distributions in both CO and H2. This was seen in the radical threshold now allow us a detailed look at the initial
bimodal correlated product state distributions from exper- phase of the roaming process leading up to the abstraction.
iment and illustrated dramatically in the dynamical calcula- The PHOFEX spectra at jCO 45 in Fig. 2 illustrate the
tions from the Bowman group. It was further confirmed in expected presence of channel 1 following excitation to 45,
the match between theory and experiment in the energy 2161, and 2143 bands in S1, an energy roughly 2500 cm1
dependence of the branching between channel 1 and chan- above the molecular transition state saddle point. How-
nel 3. We now examine the implications of the new results ever, the analogous spectra recorded at jCO 10 and 15 sug-
presented in the preceding section. gest dissociation via channel 3 (roaming) at 2161, and 2143
but not 45. The images and associated correlated state dis-
5.1. Roaming threshold tributions (Fig. 3c) clearly confirm that these low jCO prod-
ucts are indeed the result of roaming. This is quite
We can consider the roaming process in two steps: first, significant: these results allow us to bracket the occurrence
the roaming excursion, and second, the abstraction reac- of roaming in a window extending to roughly 100 cm1
tion. The H2 rovibrational distributions allowed us to below the radical threshold. These results are particularly
examine aspects of the H abstraction dynamics leading to noteworthy in light of a recent report [39] from Harding
294 S.A. Lahankar et al. / Chemical Physics 347 (2008) 288–299

a b

Fig. 6. Images (top) and translational energy distributions (bottom) acquired after excitation of: (a) para (dark line); (b) ortho (light line) transitions in
2241 and probed at jCO 15.

et al., identifying a second (or possibly multiple) saddle PES is roughly 30 cm1 below the PES De and at an out-
point on the potential energy surface using CASPT2 and of-plane angle of roughly 120° which is fully consistent
CAS+1+2+QC calculations. It is predicted to lie only with the TSR result of Harding et al. Futhermore, the con-
40–80 cm1 below the radical threshold, with a C–HR dis- tours shown are spaced 30 cm1 apart, underlining how flat
tance of 6.6 bohr and a nearly perpendicular position of the this ‘‘saddle point” is. Our PHOFEX spectra also clearly
roaming atom with respect to the HCO plane. They report support the notion of a barrier for the roaming reaction,
an imaginary frequency of 156 cm1 and extremely low real although they necessarily place it >87 cm1 below the rad-
frequencies of 22 and 83 cm1 [39,45]. They note that the ical dissociation asymptote, which is slightly below the
large amplitude, anharmonic nature of this roaming transi- lower limit suggested by Harding et al.
tion state (TSR) suggests it should really be considered a Our goal now is to understand the nature of the roaming
‘‘region” of the surface controlling the roaming reaction, excursion, its relation to TSR, and the implications of this
not a well-defined TS. Motivated by Harding’s result and for the energy dependence of the roaming pathway. In our
our PHOFEX spectra, we have examined the fitted poten- initial report, we included animations of two of the calcu-
tial in the region reported for TSR. Relevant scans of the lated trajectories: one characteristic of the roaming path-
PES are shown in Fig. 7 for several C–HR distances as a way and another representing a typical molecular
function of in-plane (h) and out-of-plane (u) angles of dissociation via channel 1. We subsequently showed that
HR with respect to HCO in its equilibrium geometry. These the roaming pathway could be effectively visualized using
scans show evidence for this barrier as a ridge surrounding a multidimensional plot of the entire trajectory showing
the abstraction region in the center of the figures. For the only the two H atoms. In Fig. 8, we show similar trajectory
scan at a C–HR distance of 6.5 bohr the barrier on the ‘‘images” to shed light on these issues. These show one
 S.A. Lahankar et al. / Chemical Physics 347 (2008) 288–299 295

Fig. 7. Contours of the roaming hydrogen at fixed RCH distance rotating around HCO in the radical equilibrium geometry. Panel (a) shows an energy
range of 33,000–34,000 cm1 with contour lines every 80 cm1 for RCH = 6.0 bohr. An extra contour line is added at 33,234 cm1 to show the approximate
threshold to hydrogen abstraction at this distance. Panels (b) (RCH = 6.614 bohr) and (c) (RCH = 7.0 bohr) show an energy range of 33,000–33,300 cm1
with contour lines at every 30 cm1. The threshold at these distances is approximately 33,210 cm1.

non-roaming event (Fig. 8a), numerous examples of roam- (Note the harmonic zero-point energy from the PES is
ing trajectories (Fig. 8b–e), and finally, one roaming trajec- 5844 cm1 and this energy must be subtracted from the
tory that yields vibrationally excited CO (Fig. 8f). total energies relative to the global minimum to make a
Accompanying each trajectory image is a plot of the two direct correspondence with the experimental energies.)
C–H distances and total potential energy at each time step These images and associated plots are rich in informa-
of the trajectory. The trajectories in Fig. 8a and b were per- tion. In the images, one can readily identify the pre-roam-
formed at 36,200 cm1 above the H2CO global minimum ing accumulation of energy in the C–HR coordinate.
while the remainder were run at 36,500 cm1, well above During roaming, the potential energy oscillates about an
the saddle point identified by Harding and coworkers. average high value due mainly to HCO vibration and
296 S.A. Lahankar et al. / Chemical Physics 347 (2008) 288–299

Fig. 8. H atom trajectory plots (left) and potential energy plots (right). The potential energy (red lines) are shown along with the C–H distances (roaming
in blue, non-roaming, black). These trajectories yield the following CO and H2 product states: (a) CO (v = 0, j = 41), H2 (v = 1, j = 7) (nonroaming);
(b) CO (v = 0, j = 6), H2 (v = 7); (c) CO (v = 0, j = 17), H2 (v = 6, j = 7); (d) CO (v = 0, j = 9), H2 (v = 6, j = 2); (e) CO (v = 0, j = 15), H2 (v = 5, j = 10);
(f) CO (v = 1, j=22), H2 (v = 6, j = 5). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this
article.)

HCO–HR orbital motion. The C–HR distance oscillates 8 bohr. The former value is in good agreement, if slightly
slightly but in the region of roaming and just prior to reac- larger than, the 6.6 bohr value of the TSR. We can also
tion it is roughly 7 bohr and in one example closer to look closely at the total potential energy in the roaming
 S.A. Lahankar et al. / Chemical Physics 347 (2008) 288–299 297

region to gain further insight. For the long roaming event just scanned across the sparse Ka = 1 states, the result is
in Fig. 8d, for example, the potential energy during the a continuum of energies at which roaming is readily possi-
excursion reaches a minimum of 33,114 cm1 and a maxi- ble. It is interesting to speculate whether the fluctuations in
mum of 36,162 cm1. The difference between these values, HCO product density of states seen in the diagram could
3000 cm1, is essentially the vibrational excitation in the give rise to fluctuations in the roaming yield. Our initial
HCO, while the difference between the maximum and the attempts to see evidence of this have not proved fruitful.
available energy, 328 cm1, must be in HCO rotation One of the challenges, as may be seen in the diagram, is
and HCO–H orbital motion. that there happen to be no states in S1 we can use to access
Fig. 8 also demonstrates another key aspect of the roam- the sparsest regions of state density in HCO.
ing behavior: the roaming process and the abstraction are
not coupled. This may be seen most clearly in Fig. 8f, in 5.2. Roaming at high energy
which two distinct roaming regions are seen, both exceed-
ing 6.6 bohr C–HR, and both showing high average poten- At excitation energies higher than the threshold of the
tial energy values. The initial roaming excursion may be triplet surface dissociation, channel 4, it becomes dominant
seen in the trajectory image as the first loop, where the tra- [12,34]. However, Fig. 4 provides clear evidence of dissoci-
jectory also becomes deep blue showing very slow H atom ation via channel 3 above the threshold of channel 4. PHO-
motion. The C–HR distance reaches 6.6 bohr and the FEX scans in this energy region (not shown here) revealed
potential energy averages 32,850 cm1 in that region. It that 35217 cm1 is the highest energy where the roaming
seems clear reaction will occur only if the roaming atom mechanism is observed. This energy is very close to the
accesses the appropriate configuration during its explora- HCO secondary dissociation threshold, where roaming
tions, and it is not simply a matter of C–H distance. could not be supported by a bound HCO species. Fig. 4
Here we have considered channel 3 in the vicinity of the also shows that even at very high energies, the internal
radical threshold, where the near dissociation must corre- energy distribution of product H2 is governed by previ-
spond to formation of HCO in its lowest quantum states. ously reported abstraction dynamics [14]. Vibrational exci-
However, we have seen that roaming also occurs above this tation in H2 is observed up to m = 10.
threshold, and grows in importance relative to channel 1.
This ‘‘roaming window” picture provides a basis for under- 5.3. Roaming leading to vibrationally excited CO
standing the nature of the roaming motion at higher ener-
gies as well. We can separate the internal degrees of Moore and coworkers have shown previously that disso-
freedom of HCO from the roaming process and the ciation of formaldehyde via channel 1 leads to vibrationally
low-frequency motions of TSR. This is essentially a vibra- excited CO (v = 1) with a yield of 15% for dissociation via
tionally adiabatic picture of the roaming dynamics. This the 2143 transition [20]. In the van Zee paper that originally
is supported by the TSR vibrational frequencies [45], 159i, suggested an alternative pathway to molecular products
22, 83, 1114 1873 and 2739 cm1, the last three of which [10], evidence was seen for this process leading to CO
may be compared to HCO frequencies of 1114, 1874 and (v = 1). This was also seen recently in the QCT calculations
2740 at the same level of theory. This close comparison from the Bowman group [38]. The data in Fig. 5 provides
clearly indicates that the saddle point represents essentially direct experimental confirmation of roaming leading to
free HCO with a loosely bound H atom. If there is more vibrationally excited CO (v = 1). Fig. 5a shows that after
energy in the radical dissociation coordinate than necessary dissociation via the 2441 band, the internal state distribu-
to form H with HCO in a given quantum state, then roam- tion of H2 formed in coincidence with CO (v = 1, j = 14)
ing will not occur and radical dissociation will be the out- peaks at around m = 6, j = 9; m = 7, j = 7; m = 8, j = 7 of
come. If the system possesses an energy in the range 0– H2. This is consistent with the roaming model, where it
100 cm1 below this threshold, then roaming motion can was shown that the H-abstraction is responsible for the
occur leading to intramolecular abstraction. Finally, if vibrational excitation of H2 [14]. Fig. 5b shows that the
the system is not within 100 cm1 of threshold for forma- internal energy distribution of H2 formed in coincidence
tion of HCO in a given quantum state, then roaming will with CO (v = 1, j = 14) following dissociation via the 2143
not occur unless there is coupling of the HCO excitation band peaks around m = 6, j = 9; m = 7, j = 3; m = 8, j = 1
into the low frequency modes. In this picture, the of H2. This is also in agreement with the roaming model
100 cm1 window may be considered to exist below each [14]. However, constraints on the available energy restrict
asymptotic HCO product state. This is schematically illus- the higher rotational excitation in H2 for dissociation via
trated in Fig. 9a and b. In Fig. 9b, we show all possible the 2143 level. Fig. 8f shows a roaming trajectory that gives
product HCO energy levels, as well as those for the more rise to vibrationally excited CO product. It is interesting to
likely Ka = 1 product states. In addition, we show a rectan- note that there is nothing particularly significant to distin-
gle representing our nominal 100 cm1 roaming ‘‘window”, guish this trajectory from the others aside from an initial
as well as the available transitions in S1 for formaldehyde unsuccessful roaming event discussed above. The average
that we use to access these various energies (Fig. 9a). It is potential energy and the oscillations in the potential energy
clear if we consider this 100 cm1 roaming window, even do not suggest any precursor in the dynamics. It is reason-
298 S.A. Lahankar et al. / Chemical Physics 347 (2008) 288–299

Fig. 9. Upper plot illustrates concept of the ‘‘roaming window” (RW) relative to the S1 levels near the radical threshold (see text). Lower plot shows
numerous levels in S1 and possible HCO product states, along with ‘‘roaming continuum” formed by considering roaming to be possible 100 cm1 below
each HCO product state.

able that CO (v = 1) formation occurs directly in the odd rotational levels implying ortho-H2, while para-formal-
abstraction step and depends on geometry and vibrational dehyde yields even rotational levels implying para-H2. This
phases at that moment. is not surprising, as nuclear spin relaxation is very slow in
general; in formaldehyde at room temperature it was mea-
5.4. Nuclear spin conservation in roaming sured by Moore et al. and found to be 0.1 s1 [18].

Previous investigations of dissociation via channel 1 6. Conclusion

have shown conservation of nuclear spin during the disso-
ciation [18]. The dissociation of ortho-formaldehyde via A comprehensive investigation of the roaming mecha-
channel 1 results in the formation of ortho-H2 and the dis- nism in formaldehyde dissociation is reported. Here all
sociation of para-formaldehyde via channel 1 result in for- possible energy regions in the formaldehyde absorption
mation of para-H2. Fig. 6 shows similarly that nuclear spin spectrum leading to dissociation via channel 3 are covered.
is largely conserved during the dissociation via channel 3, The threshold of channel 3 is between 30,240 and
since ortho-formaldehyde clearly gives rise primarily to 30,123 cm1, largely consistent with a transition state
 S.A. Lahankar et al. / Chemical Physics 347 (2008) 288–299 299

‘region’ suggested by Harding et al. The energy dependence [18] B. Schramm, D.J. Bamford, C.B. Moore, Chem. Phys. Lett. 98 (1983)
of the roaming pathway is presented in terms of a ‘‘roam- 305.
[19] D. Debarre, M. Lefebvre, M. Pealat, J.P.E. Taran, D.J. Bamford,
ing window” that views the internal excitations of HCO C.B. Moore, J. Chem. Phys. 83 (1985) 4476.
as isolated from the roaming and abstraction event. [20] D.J. Bamford, S.V. Filseth, M.F. Foltz, J.W. Hepburn, C.B. Moore,
The dissociation via channel 3 is not observed above J. Chem. Phys. 82 (1985) 3032.
35,217.4 cm1 owing to the opening of three-body dissoci- [21] K.L. Carleton, T.J. Butenhoff, C.B. Moore, J. Chem. Phys. 93 (1990)
ation processes that cannot support roaming. The dissoci- 3907.
[22] T.J. Butenhoff, K.L. Carleton, C.B. Moore, J. Chem. Phys. 92 (1990)
ation via channel 3 also leads to vibrationally excited CO 377.
and its dynamics are governed by the previously reported [23] T.J. Butenhoff, PhD, University of California, Berkeley, CA, 1990.
roaming atom model. Nuclear spin is largely conserved [24] T.J. Butenhoff, K.L. Carleton, R.D. Van Zee, C.B. Moore, J. Chem.
during the dissociation of formaldehyde via channel 3. Phys. 94 (1991) 1947.
[25] W.H. Green Jr., C.B. Moore, W.F. Polik, Annu. Rev. Phys. Chem. 43
(1992) 591.
Acknowledgements [26] R.D. Van Zee, C.D. Pibel, T.J. Butenhoff, C.B. Moore, J. Chem.
Phys. 97 (1992) 3235.
A.G.S. thanks Dr. Greg Hall for many helpful discus- [27] S.D. Chambreau, S.A. Lahankar, A.G. Suits, J. Chem. Phys. 125
sions. This work was supported by the Director, Office of (2006) 044302.
Science, Office of Basic Energy Sciences, Division of Chem- [28] M. Dupuis, W.A. Lester Jr., B.H. Lengsfield III, B. Liu, J. Chem.
Phys. 79 (1983) 6167.
ical Sciences, Geosciences and Biosciences, of the US [29] J.D. Goddard, H.F. Schaefer, J. Chem. Phys. 70 (1979) 5117.
Department of Energy under contracts DE-FG02- [30] J.D. Goddard, Y. Yamaguchi, H.F. Schaefer III, J. Chem. Phys. 75
04ER15593 (A.G.S.) and DE-FG02-97ER14782 (J.M.B.). (1981) 3459.
[31] H.B. Schlegel, S.S. Iyengar, X. Li, J.M. Millam, G.A. Voth, G.E.
References Scuseria, M.J. Frisch, J. Chem. Phys. 117 (2002) 8694.
[32] M.J. Frisch, R. Krishnan, J.A. Pople, J. Phys. Chem. 85 (1981)
1467.
[1] G.H. Dieke, G.B. Kistiakowsky, Phys. Rev. 45 (1934) 4.
[33] H.M. Yin, K. Nauta, S.H. Kable, J. Chem. Phys. 122 (2005), 194312/
[2] G.H. Dieke, G.B. Kistiakowsky, Proc. Natl. Acad. Sci. 18 (1932) 367.
1-12/10.
[3] B.A. DeGraff, J.G. Calvert, J. Am. Chem. Soc. 89 (1967) 2247.
[34] H.M. Yin, S.H. Kable, X. Zhang, J.M. Bowman, Science 311 (2006)
[4] R.D. McQuigg, PhD, Ohio State University, Columbus, OH, 1965.
1443.
[5] J.A. Pople, J.W. Sidman, J. Chem. Phys. 27 (1957) 1270.
[35] S.E. Waugh, A.C. Terentis, G.F. Metha, S.H. Kable, Proc. SPIE –
[6] R.G. Miller, E.K.C. Lee, Chem. Phys. Lett. 27 (1974) 475.
Int. Soc. Opt. Eng. 3271 (1998) 36.
[7] P.L. Houston, C.B. Moore, J. Chem. Phys. 65 (1976) 757.
[36] A.C. Terentis, P.T. Knepp, S.H. Kable, Proc. SPIE – Int. Soc. Opt.
[8] C.B. Moore, J.C. Weisshaar, Annu. Rev. Phys. Chem. 34 (1983) 525.
Eng. 2548 (1995) 328.
[9] J. Troe, J. Phys. Chem. 88 (1984) 4375.
[37] A.C. Terentis, S.E. Waugh, G.F. Metha, S.H. Kable, J. Chem. Phys.
[10] R.D. van Zee, M.F. Foltz, C.B. Moore, J. Chem. Phys. 99 (1993)
108 (1998) 3187.
1664.
[38] X. Zhang, J.L. Rheinecker, J.M. Bowman, J. Chem. Phys. 122 (2005)
[11] A.C. Terentis, S.H. Kable, Chem. Phys. Lett. 258 (1996) 626.
114313.
[12] L.R. Valachovic, M.F. Tuchler, M. Dulligan, T. Droz-Georget, M.
[39] L.B. Harding, S.J. Klippenstein, A.W. Jasper, Phys. Chem. Chem.
Zyrianov, A. Kolessov, H. Reisler, C. Wittig, J. Chem. Phys. 112
Phys. 31 (2007) 4055.
(2000) 2752.
[40] D. Townsend, M.P. Minitti, A.G. Suits, Rev. Sci. Instrum. 74 (2003)
[13] D. Townsend, S.A. Lahankar, S.K. Lee, S.D. Chambreau, A.G.
2530.
Suits, X. Zhang, J. Rheinecker, L.B. Harding, J.M. Bowman, Science
[41] W. Li, S.D. Chambreau, S.A. Lahankar, A.G. Suits, Rev. Sci.
306 (2004) 1158.
Instrum. 76 (2005), 063106/1-06/7.
[14] S.A. Lahankar, S.D. Chambreau, D. Townsend, F. Suits, J. Farnum, X.
[42] A.V. Komissarov, M.P. Minitti, A.G. Suits, G.E. Hall, J. Chem.
Zhang, J.M. Bowman, A.G. Suits, J. Chem. Phys. 125 (2006) 044303.
Phys. 124 (2006) 14303.
[15] S.A. Lahankar, S.D. Chambreau, X. Zhang, J.M. Bowman, A.G.
[43] X. Zhang, S. Zou, L.B. Harding, J.M. Bowman, J. Phys. Chem. A 108
Suits, J. Chem. Phys. 126 (2007), 044314/1-14/8.
(2004) 8980.
[16] G.D. Smith, L.T. Molina, M.J. Molina, J. Phys. Chem. A 106 (2002)
[44] D. Farnum John, X. Zhang, M. Bowman Joel, J. Chem. Phys. 126
1233.
(2007) 134305.
[17] P. Ho, D.J. Bamford, R.J. Buss, Y.T. Lee, C.B. Moore, J. Chem.
[45] L.B. Harding, Personal Communication.
Phys. 76 (1982) 3630.