Direct determination of the spin-orbit reactivity in CI(2P3/2, 2P1/2) + H2/D2/HD reactions

Article abstract of DOI:10.1063/1.1378834

Relative cross sections of the two spin orbit states in the Cl(²P)+H₂ reeaction systems was presented. Large nonadiabatic reactivity was brought about by comparing the double differential cross sections of this reaction from the two

Full text of DOI:10.1063/1.1378834

Direct determination of the spin-orbit reactivity in Cl ( 2 P 3/2 , 2 P 1/2 )+ H 2 / D 2 / HD
reactions
Feng Dong, Shih-Huang Lee, and Kopin Liu
Citation: The Journal of Chemical Physics 115, 1197 (2001); doi: 10.1063/1.1378834
View online: http://dx.doi.org/10.1063/1.1378834
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/115/3?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Spin-orbit quenching of Cl(2 P 1/2) by H2
J. Chem. Phys. 136, 124312 (2012); 10.1063/1.3697541
The investigation of spin–orbit effect for the F ( 2 P)+ HD reaction
J. Chem. Phys. 120, 6000 (2004); 10.1063/1.1650302
Cl+HD (v=1;J=1,2) reaction dynamics: Comparison between theory and experiment
J. Chem. Phys. 112, 670 (2000); 10.1063/1.480602
Energy-dependent cross sections and nonadiabatic reaction dynamics in F ( 2 P 3/2 , 2 P 1/2 )+n– H 2 →HF
(v,J)+ H
J. Chem. Phys. 111, 8404 (1999); 10.1063/1.480182
State-specific excitation function for Cl( 2 P)+H 2 (v=0,j): Effects of spin-orbit and rotational states
J. Chem. Phys. 110, 8229 (1999); 10.1063/1.478735
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:
75.102.71.33 On: Mon, 24 Nov 2014 17:06:53

JOURNAL OF CHEMICAL PHYSICS
VOLUME 115, NUMBER 3
15 JULY 2001
2
Direct determination of the spin-orbit reactivity in Cl„ P3Õ2 ,2P1Õ2…¿H2 ÕD2 ÕHD
reactions
Feng Dong,a) Shih-Huang Lee,b) and Kopin Liuc)
Institute of Atomic and Molecular Science (IAMS), P.O. Box 23-166, Academia Sinica, Taipei 106, Taiwan
͑Received 26 February 2001; accepted 18 April 2001͒
By exploiting two different Cl-beam sources and concurrently monitoring the concentrations of the
2
two reagents ͓Cl(2P3/2) and Cl ( P1/2)͔ and the H- or D-atom product, the spin-orbit specific
*
excitation functions of the title reactions were determined. The exceptionally large nonadiabatic
2
*
reactivity for Cl ( P1/2)ϩn-H2, inferred in our previous differential cross section investigation, is
now confirmed and quantified. The isotope effects for both the spin-orbit ground and excited
reagents are also elucidated. © 2001 American Institute of Physics. ͓DOI: 10.1063/1.1378834͔
I. INTRODUCTION
otherwise identical conditions, in a crossed-beam experi-
ment. Photolysis of Cl2 ͑1% seeded in H2͒ by a 355 nm laser
was used for a nearly pure Cl(2P3/2) beam. A discharge
source ͑4%–5%Cl2 seeded in He͒ was adapted to generate a
Cl(2P) beam containing both spin-orbit states with unknown
concentrations at that time. On the basis of the requirement
that the differential cross sections from either spin-orbit re-
agent must be non-negative, it was found that the reactions
with the two spin-orbit states from the discharge source
make about equal contributions to the observed dynamical
The spin-orbit effect is ubiquitous in many chemical pro-
cesses involving open-shell species. The current state of ex-
perimental work has been summarized in two comprehensive
reviews.1,2 Donovan and Husain1 examined the thermal ki-
netics of electronically excited atoms, whereas Dagdigian
and Campbell2 focused on the dynamical aspect of spin-orbit
effects. A propensity rule on spin-orbit specific reactivity has
emerged from those studies by correlating the individual
electronic fine-structure state of the reagent and product. It
was found that the electronically correlated spin-orbit state
of the reagent ͑or the product͒ always exhibits higher reac-
tivity ͑or yield͒. In other words, the adiabatically allowed
reaction pathway is generally the preferred one, and thus the
electronically nonadiabatic transition makes only a minor
contribution to the total reactivity. This rule has stood the test
of time since then, and has become a really powerful one that
allows prior prediction about spin-orbit reactivity to be made
with great confidence.
attribute. Because it is very unlikely that the discharged Cl
2
beam contains more Cl ( P1/2) than Cl(2P3/2), a larger re-
*
2
*
action cross section for Cl ( P1/2)ϩH2 than that for
Cl(2P3/2)ϩH2 was then inferred. A pressing issue is then,
*
how can one actually quantify the relative cross sections, ␴
versus ␴, for future comparison with theory and for deeper
understanding of this surprising result? Reported here is our
attempt for such a direct determination.
II. EXPERIMENT
However, ‘‘a rule is made to be broken.’’ In a recent
The experiment was performed with a rotating-source,
crossed-beam machine.4,5 The experimental approach to de-
termine the relative spin-orbit cross sections is straightfor-
ward. The same strategy of using two different sources was
previously used to unravel an intriguing energy transfer pro-
investigation of the reaction of Cl(2P)ϩH2(X1⌺ϩ)
→HCl(X1⌺ϩ)ϩH(2S) at Ecϭ5.2 kcal/mol,3 we discovered
2
*
that the spin-orbit excited Cl ( P1/2) state appears to be
more reactive to H2 than the ground state Cl(2P3/2) reaction.
Since electronically the product HCl(X1⌺ϩ)ϩH(2S) corre-
lates adiabatically with the Cl(2P3/2)ϩH2(X1⌺ϩ) reactant
through the ground state potential energy surface
6
2
˜
cess in the NCO(X ⌸)ϩHe system, and more recently in
the study of the spin-orbit reactivity of the Al(2P)ϩO2
reaction.7 For a given Cl-beam source the signal for product
yield can be expressed as ͑neglecting the uninterested con-
stants͒
(PES)12A (12⌺ϩ) in planar ͑collinear͒ geometry, this find-
Ј
ing clearly violates the conventional adiabatic correlation
rule and is in sharp contrast to what is hitherto accepted for
this particular reaction.
*
*
͑
c
Y E ͒/gϭn ␴ E ͒ϩn ␴ E ͒,
͑1͒
͑
͑
i
c
i
c
i
The surprising result of this exceptionally large nonadia-
batic reactivity was brought about by comparing the doubly
differential cross sections ͑in angles and speeds͒ of this re-
action from the two different Cl-atom beam sources, under
where Yi(Ec) is the H- or D-atom product yield at the colli-
sion energy Ec for the source i ͓either photolysis ͑p͒ or dis-
charge ͑d͔͒, and g is the initial relative velocity. The ni and
ni correspond to the initial concentrations of the Cl(2P3/2
)
*
2
*
and Cl ( P1/2) atoms in the beam from the source i. And
a͒
Present address: Department of Chemistry, University of North Carolina,
*
␴(Ec) and ␴ (Ec) denote the relative reaction cross sections
Chapel Hill, North Carolina 27599.
2
from Cl(2P3/2) and Cl ( P1/2) reagents at the collision en-
*
b͒
Present address: Department of Chemistry, National Tsing-Hua University,
ergy Ec . ͓The presence of the two 35Cl and 37Cl isotopes
leads to negligible ͑Ͻ0.6%͒ variations in collision energies.͔
Hsinchu 300, Taiwan.
Electronic mail: kpliu@gate.sinica.edu.tw
c͒
0021-9606/2001/115(3)/1197/8/$18.00
1197
© 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:
75.102.71.33 On: Mon, 24 Nov 2014 17:06:53

1198
J. Chem. Phys., Vol. 115, No. 3, 15 July 2001
Dong, Lee, and Liu
detection of the Cl(2P3/2) state͔ were used to deduce the
2
relative concentrations of the Cl(2P3/2) and Cl ( P1/2) at-
*
oms. The linearity of the VUV-intensity dependence of the
Cl(2P3/2) signal was checked by detuning the VUV-cell pres-
sures while keeping everything else the same. Similar experi-
2
*
ment has proven to be difficult for Cl ( P1/2) detection be-
cause of the extremely sensitive dependence on the phase-
matched pressure which prevents us from varying the VUV
intensity over a wide range. Nevertheless, using the same
scheme, the determination of spin-orbit populations from the
photolysis source ͑vide infra͒ is in corroboration with the
literature values.10,11 In this regard, it should also be noted
2
2
2
that the measured intensities for the P3/2→ D3/2 and P3/2
2
→ D5/2 peaks seen in Fig. 1 are in good agreement with the
reported oscillator strengths,8,9 which differ by more than
one order of magnitude.
FIG. 1. Typical REMPI signals for the two spin-orbit states of the Cl atom
from a discharge source and the D-atom product from the Cl(2P)ϩD2 re-
action. The VUV-cell conditions are also indicated at the top. Note that the
intensity ratio for the Cl(2P3/2)→Cl(2D5/2) and Cl(2P3/2)→Cl(2D3/2) tran-
sitions is about 12.5, which agrees well with the oscillator strength ratio,
12.9, found in the literature ͑Refs. 8 and 9͒.
In Eq. ͑1͒ an implicit assumption was made that only the
2
Cl(2P3/2) and Cl ( P1/2) atoms are responsible for the pro-
*
duction of the H/D products in our experiments. The justifi-
cations of this assumption are the following. First, some of
ϩ
ϩ
Ϫ
*
the potentially interfering species ͑Cl , Cl2 , e , and He ,
etc.͒ were totally removed from the experimental setup, as
discussed previously.3 Second, the flight time for the radical
beam to reach the scattering zone is around 100 ␮s, which
then eliminates many of possible metastable species, such as
*
By measuring Yi , ni , and ni from the two different sources
at the same Ec , two independent equations like Eq. ͑1͒ can
*
be set up. The desired quantities of ␴(Ec) and ␴ (Ec) can
quartet Cl and triplet Cl2. Perhaps the most compelling evi-
then be determined.
2
*
dence for assigning Cl ( P1/2) as the additional reagent from
Experimentally, what differentiates the present study
from the previous ones3,5 is the probe laser system which
now allows us to make all necessary measurements, back to
back, in one setup. The frequency-doubled output, typically
about 5 mJ/pulse over the wavelength region of our interest,
of a LD-700 dye laser was used for generating a tunable
vacuum ultraviolet ͑VUV͒ laser in a gas cell. The
(VUVϩUV) resonantly enhanced multiphoton ionization
the discharge source comes from the differential cross sec-
tion measurement. As reported previously for the Cl(2P)
ϩn-H2 reaction, the observed excess H-atom kinetic energy
release from the discharged Cl beam is entirely consistent,
within experimental accuracy of 0.2 kcal/mol, with the reac-
tion of Cl ( P1/2)ϩn-H2.3 This tight energetics consider-
2
*
ation essentially eliminates all other possibilities for interfer-
ence. Similar confirmation was also found in the differential
cross section measurements for the Cl(2P)ϩD2 studies at
two different collision energies ͑unpublished data͒.
By placing the probe laser at the signal peak, the tempo-
ral profiles of the two spin-orbit states can be monitored.
Figure 2 presents a typical result for the ClϩD2 reaction
from the discharged Cl beam. For clarity the relative inten-
͑REMPI͒ detection scheme was then used for interrogating
2
the H/D, Cl(2P3/2) and Cl ( P1/2) by simply tuning the dye
*
laser to the appropriate atomic transitions. The VUV laser
intensities were simultaneously monitored and accounted for
in the data analysis.
A. Characterization of the Cl beams
2
Detection of the Cl(2P3/2) and Cl ( P1/2) states was
sities of each species are arbitrarily scaled. As is seen, both
*
2
2
*
made by the (1ϩ1 )REMPI scheme via the P3/2
the reagent Cl ( P1/2) atom and the product D atom exhibit
Ј
2
2
2
→ D5/2,3/2 transitions near 118.87 nm and the P1/2→ D3/2
transition at 120.13 nm, respectively. Corresponding VUV
lasers were produced by the third-harmonic generation
scheme in a gas cell filled with either Xe or Kr. Figure 1
shows typical results for the discharged Cl beam. Also indi-
cated in Fig. 1 are the VUV-cell conditions. Note that 121.13
nm is near the edge of the tuning range of frequency tripling
in Kr, the optimum pressure was substantial higher. The re-
sulting VUV intensity, albeit weaker by a factor of 5, is seen
to be adequate for the present study. To monitor the Cl atoms
the mass gate was set at the m/zϭ35 ion mass channel.
Identical results were obtained for 37Cl except that the inten-
sities were reduced by a factor of 3 in accord with the natural
isotopic abundance. After accounting for the relative VUV
intensities, the literature values of the oscillator strengths8,9
for the two atomic transitions ͓a ratio of 1.16 in favor of the
a narrowly peaked profile at the same time delay between the
onset of the pulsed discharge and the probe laser, ⌬t, with a
full width at half maximum ͑FWHM͒ of about ϳ12 ␮s. The
width is substantially shorter than the gas pulse duration of
about 200 ␮s measured by both a fast ionization gauge and a
discharge current monitor. In addition to the similar peak, the
profile for Cl(2P3/2) indicates, however, another feature at
much longer ⌬t ͑not shown completely͒ which lasts for an-
other ϳ150 ␮s. It turns out that the additional longer time
feature is not due to the discharge-generated Cl(2P3/2) atom,
but, rather, arises from the VUV photochemistry of the Cl2
precursor by the probe lasers. The investigation of this VUV
spectroscopy and photodynamics will be reported in the fu-
ture. Here it suffices to say that this probe-generated
Cl(2P3/2) signal contributes no more than 5% of the total
Cl(2P3/2) signal at the typical ⌬tϭ85–90 ␮s used in the
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:
75.102.71.33 On: Mon, 24 Nov 2014 17:06:53

J. Chem. Phys., Vol. 115, No. 3, 15 July 2001
Spin-orbit reactivity
1199
*
FIG. 3. Dependence of ␴ /␴ on Sd /Sp . For further details, see the text.
2
FIG. 2. Temporal profiles of the Cl(2P3/2), Cl ( P1/2), and D atoms for the
*
ClϩD2 reaction. The probe laser wavelengths were fixed at the peaks of
their REMPI signals. The ⌬t corresponds to the time delay between the
onset of the pulsed discharge current and firing of the probe laser. Only the
early parts of the whole temporal profiles are displayed.
confined in a narrow range of the backward scattered
hemisphere.3 From these, in conjunction with the short Cl-
beam pulses and the large probe volume used in this study,
the density-to-flux correction was found to be less than 5%,
which is within the measurement uncertainties.͔ Several
back-to-back wavelength scans at fixed intersection angles
were then performed to normalize the excitation functions
for different isotopomers. The uncertainties in the relative
cross section measurements are all within Ϯ5%. In this way,
two sets of normalized apparent excitation functions were
present study, and can be readily subtracted. For reasons that
are unclear the present discharge source yields a narrow
pulse of the Cl beam only in the leading part of the gas pulse.
By way of contrast, when the same setup was used in our
study of the FϩH2 /D2 /HD reactions, a ‘‘normal’’ pulse du-
ration of the F-atom beam was found.12,13
obtained, one for the photolysis Cl beam S (E ) and the
͓
͔
p
c
other for the discharge source S (E ) .
͓
͔
Similar characterizations were also made for the pho-
tolysis source. In this case, we simply changed the sample
gases, turned off the discharged high voltage, and fired the
355 nm photolysis laser just in front of the discharge chan-
nel. In good agreement with the other photochemical
d
c
The next step is to normalize these two sets of data and
to deduce the desired reaction cross sections ͓␴(Ec) and
*
␴ (Ec)͔ from the apparent cross sections. For a given target
molecule at fixed collision energy, all three species,
2
Cl(2P3/2),Cl ( P1/2), and H or D, were monitored first for
studies,10,11 the relative branching of Cl ( P1/2) is only
2
*
*
about 0.6% of Cl(2P3/2). This provides strong support for the
the one Cl-beam source, then for the other Cl source. This
completes a cycle for data analysis. This procedure was then
repeated five to six times in a back-to-back manner to access
the statistical uncertainties. Assuming the target beam inten-
sity remains the same for each cycle, it can readily be shown
from a set of Eq. ͑1͒ that
2
Cl(2P3/2) and Cl ( P1/2) results from the discharge source.
*
The temporal profiles of the two reagents Cl(2P3/2) and
2
*
Cl ( P1/2), as well as the H/D product, are all quite short,
FWHMр6 ␮s, as anticipated for a typical radical beam gen-
erated by this method.
*
*
d
*
␴
E ͒/␴ E ͒ϭn S E ͒/n S E ͒Ϫn /n .
͑2͒
͑
͑
͑
͑
c
c
p
d
c
p c
d
d
B. Experimental procedure and data analysis
2
*
For simplicity, the small intensity of Cl ( P1/2) in the pho-
*
tolysis beam, np , is omitted here in Eq. ͑2͒, which has neg-
As in our previous report, the detection of the H/D prod-
ligible effects in the actual data analysis. As can be seen, if
such measurements were performed at several collision en-
uct was made by a (1ϩ1 )REMPI scheme via the L–␣
Ј
transition.3,5 The typical wavelength scan and the temporal
profile are shown in Figs. 1 and 2, respectively. The experi-
ments were performed in the following way. For a given
Cl-beam source, the ‘‘apparent’’ excitation function, i.e., the
translational energy dependencies of the ‘‘apparent’’ cross
sections, was first obtained for each of the target molecules:
p-H2, n-H2, D2, and HD. ͓The apparent cross section differs
in general from the ratio of Yi /g because of the density-to-
flux transformation.14 However, the kinetic energy release
for the present reaction is small ͑an endothermic reaction͒
and the product angular distributions from both sources are
*
ergies, then a plot of ␴ (Ec)/␴(Ec) vs Sd(Ec)/Sp(Ec)
*
should yield a straight line with a slope of np/nd and an
*
intercept of Ϫnp/nd . And this relationship should hold even
for different target molecules provided that their apparent
cross sections have already been properly normalized. Figure
3 shows the results, in which the closed squares are for the
reaction with D2 at five different Ec’s and the two open
squares are for that with n-H2. The error limits are Ϯ2 stan-
dard deviations from error propagations. It is gratifying to
see that Eq. ͑2͒ holds within the experimental uncertainties.
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:
75.102.71.33 On: Mon, 24 Nov 2014 17:06:53

1200
J. Chem. Phys., Vol. 115, No. 3, 15 July 2001
Dong, Lee, and Liu
Several points are interesting to note. First, the magni-
*
tude of the ratio ␴ /␴ is quite significant, indicative that
*
2
Cl ( P1/2) atom is indeed remarkably reactive. Second, the
Cl(2P) beam from the discharge source is about 3.5 times
more intense than the photolysis beam. This number is not
very different from the ratio of the initial concentrations of
the Cl2 precursor of the two sources, suggesting that the two
methods have about the same efficiency in generating a Cl
beam from the Cl2 molecule. Third, for the discharged Cl
2
*
beam, the concentration of the spin-orbit excited Cl ( P1/2
)
state is substantial, about 50% of the ground state, which is
exactly the ratio of the electronic degeneracy factors. Previ-
ously, we used the translational temperature of the beam
2
*
͑ϳ600 K͒ to guess the Cl ( P1/2) population in the dis-
charged beam.3 Apparently, the spin-orbit population does
not equilibrate with the translational degree of freedom dur-
ing supersonic expansion. Such a guess underestimates its
concentration by as much as a factor of 3, which should
serve as a cautious note.
III. RESULTS AND DISCUSSION
A. Spin-orbit specific excitation functions
Because of the calibration procedure given in detail
above, all excitation functions presented in this study have
been normalized to one another. Hence, only a single scaling
factor will be needed for all isotopomers to obtain the abso-
lute cross sections and to compare with future theory. The
apparent excitation functions of the Cl(2P)ϩD2 reaction
from the two Cl-atom sources are presented in the upper
panel of Fig. 4. The lines are empirical fits to guide the eye.
*
*
Based on the best-fitted values of np /nd and nd /nd obtained
*
from Fig. 3, the relative cross sections, ␴(Ec) and ␴ (Ec),
FIG. 4. Results for the Cl(2P)ϩD2 reaction. The normalized apparent ex-
citation functions from the two different Cl sources are shown in the upper
were then deduced and are shown in the lower panel. The
error bars represent Ϯ2 standard deviations. The threshold
for the Cl ( P3/2)ϩD2 reaction appears around 5 kcal/mol.
For the reaction with Cl ( P1/2), the onset becomes coun-
*
panel. The relative excitation functions of ␴ (Ec) and ␴(Ec) deduced are
2
*
shown in the lower panel. The error bars represent 2 standard deviations.
2
*
terintuitively higher by about 1 kcal/mol, despite the fact that
it has an additional 2.5 kcal/mol from the spin-orbit excited
energy and that the nonadiabatic transition to the ground re-
active surface is expected to occur in the entrance valley. As
Cl(2P3/2)ϩD2 reaction indicative of the zero-point-energy
effect and/or the tunneling effect. Because our D2 beam con-
sists mostly of the ground rotational state and the n-H2 beam
is dominated by the jϭ1 state,13,15 one may wonder if the
*
a result, ␴(Ec)Ͼ␴ (Ec) near the threshold region. How-
*
ever, the relative reactivity reverses at higher energies be-
differences in threshold behavior of ␴ vs ␴ for these two
*
cause the ␴ (Ec) exhibits a steeper rise than ␴(Ec). At Ec
isotopic reagents arise from the different initial rotational
states.
2
*
ϳ9 kcal/mol, the nonadiabatic reactivity of Cl ( P1/2) be-
comes nearly twice that of the adiabatic one from
Figure 6 plots the results for Cl(2P)ϩp-H2. As is seen,
2
*
*
Cl ( P3/2)!
the thresholds for ␴ and ␴ are still about the same, just like
Similarly, the results for Cl(2P)ϩn-H2 are presented in
*
the case for n-H2. Hence, the delayed onset of ␴ (Ec)
2
2
*
Fig. 5. Once again the reactivity of Cl ( P1/2) is higher than
that of Cl(2P3/2). Previously, from the differential cross sec-
tion analysis of this reaction we found, totally independent of
*
shown in Fig. 4 for the Cl ( P1/2)ϩD2 reaction does not
originate from the initial j-state selection. The thresholds for
the p-H2 reaction appear to be slightly lower than the reac-
tions with n-H2. This aspect, along with the relative magni-
tudes of Figs. 5 and 6, has been the subject of a previous
report.5 The conclusion that the rotation of the reagent has a
positive impact on enhancing the reactivity is opposite to the
theoretical prediction based on the G3 PES, but in line with
the quasiclassic trajectory ͑QCT͒ results16 on a more recent
ab initio Bian-Werner ͑BW͒ PES.17,18
the present work, that the two reagents Cl(2P3/2) and
2
*
Cl ( P1/2) from the discharge source make about equal con-
tributions to the total reactivity.3 Based on the measured nd
*
*
and nd in this work, it then implies ␴ Ϸ2␴ at Ec
ϭ5.2 kcal/mol, which is in good accordance with the present
integral cross section measurement, ␴ Ϸ1.8␴. However, in
*
this case the thresholds for both spin-orbit states seem about
the same, ϳ3.8 kcal/mol, which is lower than that for the
Previously, based on the results of the photolyzed-Cl
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:
75.102.71.33 On: Mon, 24 Nov 2014 17:06:53

J. Chem. Phys., Vol. 115, No. 3, 15 July 2001
Spin-orbit reactivity
1201
FIG. 6. Excitation functions for the Cl(2P)ϩp-H2 reaction.
FIG. 5. Excitation functions for the Cl(2P)ϩn-H2 reaction. Note that the
results were normalized to the ClϩD2 reaction shown in Fig. 4, and to all
other isotopomers in Figs. 6–9.
qualitative conclusion drawn from our previous analysis.5
Finally, the results of the two isotopic product channels
for the Cl(2P)ϩHD reaction are plotted in Figs. 8 and 9. We
first note that similar behavior to that in the Cl(2P)ϩD2
ϩn-H2 /p-H2 reactions, the rotationally specific excitation
functions for Cl(2P3/2)ϩH2(jϭ0 or 1) were deduced.5 At
that time, however, the relative amounts of the two spin-orbit
states in the discharge beam were unknown. Thus, similar
quantities for the spin-orbit excited reagent could not be ob-
tained, and only a qualitative trend was inferred. It is instruc-
tive to quantify the complete state-specific excitation func-
tions now. Using the same scheme as before, i.e., assuming
reaction, i.e., a delayed onset for the reaction with
2
*
Cl ( P1/2) and a very significant nonadiabatic reactivity at
higher collision energies, prevails for both isotope channels.
Second, the relative scales for reaction cross sections of the
two isotope channels are very different. Such a large dispar-
ity in branching was noted previously for the ground spin-
orbit reagent and could be traced back to the long-range ste-
reodynamics arising from the anisotropic van der Waals
interactions in the entrance valley.19 It is interesting to note
that a similar, in fact even stronger, preference in isotope
channel was revealed here for the spin-orbit excited reagent.
In terms of the kinetics isotope effects the usual trend of
*
*
that ␴1Ϸ␴ and ␴ Ϸ␴ , the desired quantities can be de-
2
1
2
duced from the lower panels of Figs. 5 and 6, and are pre-
sented in Fig. 7. The upper panel gives the individual exci-
*
tation function, ␴ and ␴ , whereas the lower panel displays
j
j
the ratios of the spin-orbit reactivities for the rotationally
*
*
state-selective reagent. Clearly, ␴ /␴0ӷ␴ /␴ ͑or ␴1 /␴
1
0
0
1
␴
H Ͼ␴HDϾ␴ is observed for both spin-orbit reagents, al-
D
2 2
*
*
ӷ␴1 /␴ ͒ near the threshold region, but as the collision en-
0
though the relative magnitudes vary with the collision ener-
gies because of different threshold behaviors.
*
*
ergy increases, the ratio ␴0 /␴0(␴1 /␴1) decreases ͑in-
creases͒ gradually and both approach a value of about 2 at
higher energies. In other words, the relative cross section of
*
B. Comparisons with other related experiments
2
Cl ( P1/2) to Cl(2P3/2) is greater for a nonrotating H2 re-
agent than for a rotating one, and this preference is even
more pronounced at low collision energies. This confirms the
The most significant contribution of this work is the con-
firmation and quantification of unusually large nonadiabatic
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:
75.102.71.33 On: Mon, 24 Nov 2014 17:06:53

1202
J. Chem. Phys., Vol. 115, No. 3, 15 July 2001
Dong, Lee, and Liu
FIG. 7. Upper panel: Complete state-selected excitation functions, deduced
from the empirical fits shown in Figs. 5 and 6 for the Cl(2P3/2 ,2P1/2
)
ϩH2(jϭ0,1) reactions. The propensities of rotationally state-specific spin-
orbit reactivity for jϭ0 and 1 are shown in the lower panel. See the text for
details.
FIG. 8. Excitation functions for Cl(2P)ϩHD→DClϩH.
*
reactivity from the spin-orbit excited Cl atom, which was
discovered in our previous differential cross section
measurements.3 In view of the fact that the thermal popula-
tion of the spin-orbit excited state varies from ϳ0.7% at
room temperature to ϳ25% at 3000 K, its reactivity could
have a significant effect on the temperature dependence of
thermal rate constants. Because this conclusion is so unex-
pected, it is worth examining it more closely by comparing it
with several other related studies and commenting on some
controversy in the literature.
molecule reaction or a neutral reaction via the harpooning
mechanism, and is not applicable to the present case.
In a recent hot-atom experiment the absolute reactive
cross sections for the reverse reaction, HϩHCl
→H2ϩCl(2P3/2) and Cl ( P1/2), were determined for Ec
2
*
ϭ1.0–1.75 eV.22,23 It was found that the branching for the
2
*
Cl ( P1/2) product yield increases from ϳ6% to ϳ20% with
an increase in collision energies. Despite the large difference
in collision energies in the two experiments, this hot-atom
result, at first glance, seems contradictory to the conclusion
of the present study. However, the product state distribution
reported previously provides a hint for resolving this discrep-
ancy. As shown in Fig. 6 of Ref. 3, the HCl products from
To the best of our knowledge the only previous reaction
that violates the conventional spin-orbit propensity is the iso-
electronic ion–molecule reaction of Arϩ(2P3/2 ,2P1/2
)
ϩH2͑D2͒→ArHϩ͑ArDϩ͒ϩH͑D͒.20 It was found that the ex-
cited Arϩ(2P1/2) atom is more reactive than the ground
Arϩ(2P3/2) atom by a factor of ϳ1.4. To interpret this in-
triguing observation, a mechanism invoking the crossings
and transitions of the Arϩ(2P3/2)ϩH2 and Arϩ(2P1/2)ϩH2
surfaces with the ArϩHϩ2 PES was proposed. Thus, the re-
action proceeds through charge transfer followed by an
atom-exchange process. No such propensity was observed
for the analogous reaction with Krϩ(2P3/2 ,2P1/2),21 sugges-
tive of the vibronic enhancement in surface crossings. In any
event, this charge transfer mechanism is unique to an ion–
the Cl(2P3/2)ϩH2 reaction are formed with little rotational
2
*
excitation ( j ϳ4), whereas those from the Cl ( P
)
1/2
Ј
͗ ͘
ϩH reaction are mostly in the higher j states ( j ϳ9). The
Ј
͗ ͘
2
reaction cell in the hot-atom experiment is at room tempera-
ture; that is, only the low j states of HCl are present. Accord-
ing to the principle of microscopic reversibility,24 it is thus
2
not surprising to see more Cl(2P3/2) than Cl ( P1/2) prod-
*
ucts in the reverse reaction for a HCl reagent with little ro-
tational energy. If the same hot-atom experiment were per-
formed with highly rotationally excited HCl reagents, one
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:
75.102.71.33 On: Mon, 24 Nov 2014 17:06:53

J. Chem. Phys., Vol. 115, No. 3, 15 July 2001
Spin-orbit reactivity
1203
2
*
ties of the present reaction with that of the F ( P1/2) atom.
The latter reaction has received much renewed attention over
the past few years.30 In particular, by measuring the HF(j )
Ј
product state distribution for ϭ3 near threshold, Nesbitt
v
Ј
and his co-workers reported excess product yields thus pro-
viding experimental evidence for the nonadiabatic reactivity
of F ( P1/2) in the reaction of Fϩn-H2.31 At first this con-
2
*
clusion seems to be at variance with the earlier crossed-beam
scattering results in which no evidence for a finite contribu-
tion from F ( P1/2) was found.30 Nevertheless, our recent
2
*
investigation of the F(2P)ϩHD reaction suggested that there
is actually no significant conflict among any of the
experiments.13 In that study, a contribution of a few percent
2
*
͑Շ5%͒ from F ( P1/2) to the total reactivity was clearly es-
tablished over a wide range of initial collision energies,
ϳ0.3–4.5 kcal/mol. Assuming the spin-orbit populations for
the discharged F beam behave the same as the present dis-
charged Cl source, i.e., according to their degeneracy factors,
one has nFϷ2nF . This then leads to the conclusion that the
*
2
*
reactive cross section for F ( P1/2)ϩHD is no more than
10% that for F(2P3/2)ϩHD, which corroborates with a most
recent, full QM multisurface calculation for the F(2P)ϩH2
reaction.32,33 Hence, although the nonadiabatic reactivity of
2
*
F ( P1/2) toward HD is finite, it apparently plays only a
minor role in the total reactivity. This conclusion is entirely
consistent with that expected from the adiabatic correlation
rule.
By way of contrast, moving from the F atom to the Cl
2
*
atom, what reported here is the opposite: Cl ( P1/2) be-
comes more reactive than Cl(2P3/2), namely, the nonadia-
batic pathway is more facile than the adiabatic one. As
pointed out in our previous report,5 it is instructive to take an
alternative view which is based on the absolute cross sec-
tions. The reaction of FϩH2 is fast, with a typical cross sec-
tion of ϳ3 Å2.34,35 Assuming the reactive cross section for
FIG. 9. Excitation functions for Cl(2P)ϩHD→HClϩD.
*
*
F ϩH2, in analogy to the F ϩHD reaction, is about 10%,
thus it is ϳ0.3 Å2. On the other hand, the reaction of ClϩH2
is much slower. The QCT calculation based on the BW2 PES
yields about 0.15 Å2 for the Cl(2P3/2)ϩH2 reaction at Ec
would then observe the opposite, i.e., more products in the
2
*
spin-orbit excited Cl ( P1/2) state than in the ground state.
2
*
Deactivation of Cl ( P1/2) by H2 and D2 has also been
investigated kinetically.25 The relaxation rate constants are
ϭ5.3 kcal/mol.16 Hence, a factor of 2 enhancement in reac-
tivity for Cl ( P1/2), as reported here, will give ␴
Ӎ0.3 Å , which is about the same as that of F ϩH2 in ab-
solute magnitude. Hence, the exceptionally large contribu-
large, ϳ6ϫ10Ϫ11 and 1.5ϫ10Ϫ11 cm3/molecule s for H2 and
2
*
*
2
2
*
*
D2, respectively. Clearly, the deactivation of Cl ( P1/2) is
dominated by a physical quenching process to the ground
Cl(2P3/2) state. In the literature, this was often taken as evi-
tion from Cl to the overall reactivity does not necessarily
*
2
*
dence that chemical reaction between Cl ( P1/2) and H2 /D2
imply this reaction system exhibits unusually large nonadia-
batic couplings among the PESs. Rather, it could just reflect
the relatively small cross section of the ground state Cl atom.
In other words, it is a choice of reference. Nevertheless, what
is relevant for a given reaction is if the nonadiabatic multi-
surface transition, or the spin-orbit excited species which can
react only through the nonadiabatic transition in these cases,
makes a significant contribution to its kinetics and dynamical
behavior. The adiabatic correlation rule tells us that most
chemical reactions prefer the electronically adiabatic path-
way. In this sense, the dominance of nonadiabatic reactivity
observed for the Cl(2P)ϩhydrogen molecule reaction is par-
ticularly significant and intriguing. Very recently a crossed-
beam study of the ClϩD2 reaction dynamics was reported.36
Compared with the concurrent QCT calculations on the BW-
is unlikely, and that it will not contribute to the thermal ki-
netics of the Cl–H2 system. In our opinion this is not a sound
argument. Recall that the room temperature rate constant for
the ClϩH2 reaction is about 1.6ϫ10Ϫ14 cm3/molecules,26
which is more than three orders of magnitude slower than the
2
*
Cl ( P1/2) deactivation rate. Thus, if even only a small frac-
2
*
tion of the relaxations of Cl ( P1/2) by H2 is due to chemical
reaction, its reactivity can be significant compared to the
ground state reaction rate. A similar situation exists for the
analogous reaction Cl(2P)ϩCH4,25,27–29 for which the room
temperature rate constant is 1.0ϫ10Ϫ13 cm3/molecules and
2
*
the total quenching rate constant of Cl ( P1/2) by CH4 has
been measured to be about 2ϫ10Ϫ11 cm3/molecules.
Finally, it is worth to contrast the nonadiabatic reactivi-
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:
75.102.71.33 On: Mon, 24 Nov 2014 17:06:53

1204
J. Chem. Phys., Vol. 115, No. 3, 15 July 2001
Dong, Lee, and Liu
PES, good general agreement was obtained; but a subtle dis-
9 M. A. A. Clyne and W. S. Nip, J. Chem. Soc., Faraday Trans. 2 73, 1308
͑1977͒.
crepancy in the sideways scatterings was also noted. The
10 Y. Matsumi, K. Tonokura, and M. Kawasaki, J. Chem. Phys. 97, 1065
͑1992͒.
2
*
neglect of the Cl ( P1/2) reactivity in the theoretical treat-
11 P. C. Samartzis, B. L. G. Bakker, T. P. Rakitzis, D. H. Parker, and T. N.
Kitsopoulos, J. Chem. Phys. 110, 5201 ͑1999͒.
ment was conjectured to be the origin of the discrepancy.
Qualitatively, this is consistent with our findings.
12 R. T. Skodje, D. Skouteris, D. E. Manolopoulos, S.-H. Lee, F. Dong, and
K. Liu, J. Chem. Phys. 112, 4536 ͑2000͒.
IV. CONCLUSION
13 F. Dong, S.-H. Lee, and K. Liu, J. Chem. Phys. 113, 3633 ͑2000͒.
14 D. M. Sonnenfroh and K. Liu, Chem. Phys. Lett. 176, 183 ͑1991͒.
15 Y.-T. Hsu, J.-H. Wang, and K. Liu, J. Chem. Phys. 107, 2351 ͑1997͒.
16 D. E. Manolopoulos, F. J. Aoiz, and L. Banares ͑private communications͒.
17 W. Bian and H.-J. Werner, J. Chem. Phys. 112, 220 ͑2000͒.
18 U. Manthe, W. Bian, and H. J. Werner, Chem. Phys. Lett. 313, 647 ͑1999͒.
19 D. Skouteris, D. E. Manolopoulos, W. Bian, H.-J. Werner, L.-H. Lai, and
K. Liu, Science 286, 1713 ͑1999͒.
A direct determination of the relative cross sections of
the two spin-orbit states in the Cl(2P)ϩH2 reaction systems
is presented. An exceptionally large nonadiabatic reactivity,
which was inferred from our previous differential cross sec-
tion measurement, has now been firmly established. This
finding is in sharp contrast to all other known neutral bimo-
lecular reactions, and is in violation of the conventional adia-
batic correlation rule. A predictive, conceptual understanding
of this unexpected phenomenon remains a theoretical chal-
lenge. A high quality, ab initio result of both the ground and
the excited PESs for this reaction will soon become
available.37 And the exact multisurface quantum mechanical
dynamical calculation has also become feasible recently.32,33
The results presented in this work will then serve as a strin-
gent measure for future comparison. Better understanding of
this particular reaction will, we believe, provide a deeper
appreciation of the role of spin-orbit effects in chemical re-
actions in general.
20 K. Tanaka, J. Durup, T. Kato, and I. Koyano, J. Chem. Phys. 73, 586
͑1980͒; 74, 5561 ͑1981͒.
21 K. M. Ervin and P. B. Armentrout, J. Chem. Phys. 85, 6380 ͑1986͒.
22 R. A. Brownsword, C. Kappel, P. Schmiechen, H. P. Upadhyaya, and
H.-R. Volpp, Chem. Phys. Lett. 289, 241 ͑1998͒.
23 F. J. Aoiz, L. Banares, T. Bohm et al., J. Phys. Chem. A 104, 10452
͑2000͒.
24 R. D. Levine and R. B. Bernstein, Molecular Reaction Dynamics and
Chemical Reactivity ͑Oxford University Press, New York, 1987͒.
25 A. I. Chichinin, J. Chem. Phys. 112, 3772 ͑2000͒, and references therein.
26 S. S. Kumaran, K. P. Lim, and J. V. Michael, J. Chem. Phys. 101, 9487
͑1994͒.
27 A. R. Ravishankara and P. H. Wine, J. Chem. Phys. 72, 25 ͑1980͒.
28 G. S. Tyndall, J. J. Orlando, and C. S. Kegley-Owen, J. Chem. Soc.,
Faraday Trans. 91, 3055 ͑1995͒.
29 Y. Matsumi, K. Izumi, V. Skorokhodov, M. Kawasaki, and N. Tanaka, J.
Phys. Chem. A 101, 1216 ͑1997͒.
ACKNOWLEDGMENTS
30 K. Liu, Annu. Rev. Phys. Chem. 52, 139 ͑2001͒, and references therein.
31 S. A. Nizkorodov, W. W. Harper, W. B. Chapman, B. W. Blackmon, and
D. J. Nesbitt, J. Chem. Phys. 111, 8404 ͑1999͒.
This work was supported by the National Science Coun-
cil of Taiwan and by the China Petroleum Corp.
32 M. H. Alexander, H. J. Werner, and D. E. Manolopoulos, J. Chem. Phys.
109, 5710 ͑1998͒.
33 M. H. Alexander, D. E. Manolopoulos, and H. J. Werner, J. Chem. Phys.
113, 11084 ͑2000͒.
1 R. J. Donovan and D. Husain, Chem. Rev. 70, 489 ͑1970͒.
2 P. J. Dagdigian and M. L. Campbell, Chem. Rev. 87, 1 ͑1987͒.
3 S.-H. Lee and K. Liu, J. Chem. Phys. 111, 6253 ͑1999͒.
4 R. G. Macdonald and K. Liu, J. Chem. Phys. 91, 821 ͑1989͒.
5 S.-H. Lee, L.-H. Lai, K. Liu, and H. Chang, J. Chem. Phys. 110, 8229
͑1999͒.
34 D. E. Manolopoulos, J. Chem. Soc., Faraday Trans. 93, 673 ͑1997͒.
35 F. J. Aoiz, L. Banares, and V. J. Herrero, J. Chem. Soc., Faraday Trans. 94,
2483 ͑1998͒.
36 N. Balucani, L. Cartechini, P. Casavecchia, G. G. Volpi, F. J. Aoiz, L.
Banares, M. Menendez, W. Bian, and H.-J. Werner, Chem. Phys. Lett. 328,
500 ͑2000͒.
6 R. G. Macdonald and K. Liu, J. Chem. Phys. 98, 3716 ͑1993͒.
7 C. Naulin and M. Costes, Chem. Phys. Lett. 310, 231 ͑1999͒.
8 G. M. Lawrence, Astrophys. J. 148, 261 ͑1967͒.
37 H.-J. Werner ͑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:
75.102.71.33 On: Mon, 24 Nov 2014 17:06:53