Pressure dependence and metastable state formation in the photolysis of dichlorine monoxide (Cl2O)

Article abstract of DOI:10.1063/1.471109

The photolysis of dichlorine monoxide (Cl₂O) is investigated using broadband photolysis source with time-resolved absorption spectroscopy. Time-dependent concentrations of reactants and products are measured by long-path uv absorption spectrosc

Full text of DOI:10.1063/1.471109

Pressure dependence and metastable state formation in the photolysis of dichlorine
monoxide (Cl2O)
Scott L. Nickolaisen, Charles E. Miller, Stanley P. Sander, Michael R. Hand, Ian H. Williams, and Joseph S.
Francisco
Citation: The Journal of Chemical Physics 104, 2857 (1996); doi: 10.1063/1.471109
View online: http://dx.doi.org/10.1063/1.471109
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/104/8?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Statetostate dissociation of OClO(Ã2 A 2ν1,0,0)→ClO(X 2ΠΩ,v,J)+O(3 P)
J. Chem. Phys. 104, 2883 (1996); 10.1063/1.471649
193 nm laser photofragmentation timeofflight mass spectrometric study of HSCH2CH2SH
J. Chem. Phys. 104, 130 (1996); 10.1063/1.470883
Predissociation dynamics of the O2 B 3Σ− u state: Vibrational state dependence of the product finestructure
distribution
J. Chem. Phys. 103, 2495 (1995); 10.1063/1.469671
Experimental and theoretical studies of isomeric CH3S2 and CH3S+ 2
J. Chem. Phys. 100, 4870 (1994); 10.1063/1.467277
Photolysis of MatrixIsolated Dichlorine Monoxide: Infrared Spectra of ClClO and (ClO)2
J. Chem. Phys. 46, 4481 (1967); 10.1063/1.1840572
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.42.202.150 On: Mon, 24 Nov 2014 05:04:06

Pressure dependence and metastable state formation in the photolysis
of dichlorine monoxide (Cl2O)
Scott L. Nickolaisen,a) Charles E. Miller, and Stanley P. Sanderb)
Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena,
California 91109
Michael R. Hand and Ian H. Williams
School of Chemistry, University of Bath, Claverton Down, Bath BA2 7AY, United Kingdom
Joseph S. Francisco
Department of Chemistry, Purdue University, W. Lafayette, Indiana 47907-1397
͑Received 25 July 1995; accepted 15 November 1995͒
The photodissociation of dichlorine monoxide ͑Cl2O͒ was studied using broadband flash photolysis
to investigate the influence of variations in the photolysis wavelength domain, bath gas pressure and
bath gas identity on the yield and temporal dependence of the ClO product. ClO yields were
independent of bath gas pressure when the photolysis spectral band extended to 200 nm ͑quartz
cutoff͒ but for photolysis restricted to wavelengths longer than about 250 nm, ClO yields decreased
with increasing bath gas pressure and there was a pressure-dependent delay in the formation of ClO.
Under these conditions, a weak, highly structured absorption spectrum was observed in the range
16 600–26 000 cmϪ1 with a lifetime on the order of 500 ms. A portion of the spectrum could be
analyzed ͑22 000–26 000 cmϪ1͒ which showed progressions having differences of 283, 443, and
505 cmϪ1. Ab initio calculations were performed to evaluate vertical excitation energies and
1
oscillator strengths from the lowest-energy singlet (X A1) or triplet (1 3B1) states to various
excited states. The calculations indicated that the 2 3A2←1 3B1 transition has an unusually large
oscillator strength. The transition energy, 3.05 eV, is consistent with the observed metastable
spectrum. The observed pressure dependence of ClO formation could be modeled using a
mechanism which assumed that Cl2O excitation at wavelengths longer than about 300 nm leads to
rapid intersystem crossing to two metastable states in the triplet manifold. These states undergo
competitive dissociation to ClOϩCl and collisional relaxation to the ground state. The dynamics of
Cl2O may serve as a model for other molecules of importance in the earth’s lower stratosphere such
as ClONO2 where filtering of the solar spectrum by ozone restricts photolysis to the weak tail of the
absorption continuum. © 1996 American Institute of Physics. ͓S0021-9606͑96͒01608-3͔
I. INTRODUCTION
it is important to quantify the wavelength dependences of the
overall quantum yields for all photolytic channels.
Early studies of Cl2O photodissociation showed that ir-
radiation at wavelengths below 300 nm leads primarily to
formation of ClOϩCl, with minor channels leading to O(3P)
or O(1D) by the reactions
The photodissociation of chlorine oxides has received
considerable attention in recent years due to their involve-
ment in stratospheric ozone depletion by chlorofluorocar-
bons. Cl2O ͑chlorine monoxide or dichlorine monoxide͒ is
the anhydride of an important chlorine reservoir species in
the stratosphere, HOCl, but does not itself play a significant
role in atmospheric chemistry. There is, however, interest in
understanding the kinetics and photodissociation of Cl2O.
Because it is used extensively as a ClO source through direct
photolysis,
3
Cl Oϩh␯→2ClϩO P͒
␭threshϭ292 nm,
͑
2
3
Cl Oϩh␯→Cl ϩO P͒
␭
threshϭ710 nm,
threshϭ335 nm.
͑
2
2
1
Cl Oϩh␯→Cl ϩO D͒
␭
͑
2
2
Finkelnburg et al. and Schumacher and Townend used
steady-state photolysis sources combined with measurements
of the Cl2 quantum yield to deduce the primary quantum
yields.1,2 Later flash photolysis studies of Cl2O employed the
direct detection of ClO.3,4 Sander and Friedl5 observed the
formation of BrO in the broadband photolysis of Cl2O–Br2
systems at wavelengths longer than 180 nm. In this experi-
ment, the only likely pathway for the formation of BrO was
the O(3P)ϩBr2 reaction, with the O(3P) being produced by
Cl2O photolysis. The measured O(3P) quantum yield over
the photolysis band was 25%Ϯ5%. Recently, Nelson et al.6
have employed photofragment translational energy spectros-
Cl2Oϩh␯→ClOϩCl
␭threshϭ840 nm
and by the rapid reaction with atomic chlorine,
Cl2ϩh␯→2Cl,
ClϩCl2O→ClOϩCl2,
a͒
Permanent address: Department of Chemistry and Biochemistry, California
State University, Los Angeles, 5151 State University Drive, Los Angeles,
CA 90032.
b͒
Author to whom correspondence should be addressed.
J. Chem. Phys. 104 (8), 22 February 1996
0021-9606/96/104(8)/2857/12/$10.00
© 1996 American Institute of Physics
2857
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.42.202.150 On: Mon, 24 Nov 2014 05:04:06

2858
Nickolaisen et al.: Photolysis of dichlorine monoxide
nm. At wavelengths longer than this cutoff point, the absorp-
tion cross sections of many temporary reservoir species are
small due to weak Franck–Condon overlap or forbidden
electronic transitions. Nevertheless, photolysis is the most
important loss process for these molecules in the lower
stratosphere. Results of photodissociation studies carried out
at shorter wavelengths, or under different conditions of tem-
perature and pressure than those encountered in the lower
stratosphere may not be applicable to the real atmosphere.
Errors in the measurement of effective quantum yields will
have a significant impact on the ability of atmospheric mod-
els to correctly predict the abundance and partitioning of
trace species, and therefore affect calculations of strato-
spheric ozone loss. As indicated above, while Cl2O is not a
molecule that is directly relevant to atmospheric chemistry,
studies of its photodissociation at long wavelengths may
serve as a paradigm for the understanding of the photolysis
of stratospheric temporary reservoir species.
One approach to the investigation of photolysis pro-
cesses in weakly absorbing bands is to use a broadband pho-
tolysis source such as a flashlamp coupled with analysis by
time-resolved absorption spectroscopy. While a flashlamp
lacks the wavelength specificity to initiate photolysis from a
well-defined initial state, it can provide a large photolysis
flux in the region of low absorption cross section of the
photolyte. We have used this approach to study the photoly-
sis of Cl2O. Photolysis filters were used to isolate photolysis
spectral regions from 180 nm through the visible. The yields
and time-dependent concentrations of ClO were measured as
a function of buffer gas density using an optical multichannel
analyzer ͑OMA͒. The spectrum of a new intermediate was
identified in the photolysis of Cl2O at wavelengths longer
than 300 nm. Ab initio calculations have been carried out to
determine the energy ordering and oscillator strengths for
transitions in the singlet and triplet manifolds of Cl2O using
the complete-active-space self-consistent-field ͑CASSCF͒
and multireference configuration-interaction ͑MRCI͒ meth-
ods for the singlet and triplet ground and excited states, re-
spectively.
FIG. 1. Cl2O absorption cross sections vs wavelength. Bars indicate excita-
tion regions for photolysis cell/filter solution combinations discussed in the
text.
copy in a molecular beam to study Cl2O photodissociation
mechanisms at 308, 248, and 193 nm. This work showed that
primary Cl–O bond fission occurs at all three wavelengths.
At 248 and 193 nm the ClO fragment possessed sufficient
internal energy to undergo secondary dissociation and at 193
nm there was evidence for concerted Cl2 elimination path-
ways on both the singlet and triplet surfaces. While this ex-
periment provided considerable information on the dynamics
of the primary and secondary reaction channels, it did not
probe the photolysis in the weak tail of the Cl2O absorption
continuum beyond 308 nm or effects due to collisions.
Chichinin7 studied the 248 nm photolysis of Cl2O using time
resolved laser magnetic resonance to measure the extent of
spin–orbit excitation of nascent Cl(2PJ). The atomic chlo-
rine was found to be predominantly ͑82%͒ in the ground
(2P3/2) state at this wavelength but quantum yields at longer
wavelengths and the effect of buffer gas collisions were not
examined.
Cl2O is characterized by a continuous absorption spec-
trum with maxima at 530, 410, 256, and 171 nm, and a
vibrationally resolved Rydberg series with an absorption
maximum at 161.1 nm ͑Fig. 1͒.8–11 The electronic states re-
sponsible for the observed ultraviolet absorption bands have
not been determined experimentally or theoretically. Because
the Cl2O absorption cross sections decrease rapidly to values
less than 10Ϫ20 molecule cmϪ2 beyond 350 nm direct prob-
ing of the photodissociation dynamics of the visible band
systems is quite difficult due to the small fractional decom-
position of the photolyte.
Photodissociation in the weak absorption tail is a prob-
lem that is very important for stratospheric and tropospheric
chemistry. Species that are important temporary reservoirs of
odd hydrogen, nitrogen, chlorine, and bromine such as
ClONO2 and HONO2 are most abundant in the lower strato-
sphere below 30 km. In this region of the atmosphere, ozone
absorption removes virtually all solar radiation below 290
II. EXPERIMENT
Cl2O was prepared by the method of Cady12 in which a
mixture of 10%–20% Cl2 in helium was slowly flowed over
dried HgO at room temperature and at a total pressure of 250
Torr. The effluent gas was collected in a trap at 161 K. Cl2O
was purified of residual Cl2 by vacuum distillation at 196 K.
The purity of the sample was checked by uv absorption spec-
troscopy. The only detectable impurity was OClO at 0.2 pptv
with no discernible Cl2 impurity. Cl2O was introduced into
the flash photolysis apparatus by flowing UHP helium
͑Matheson͒ through the liquid sample at 196 K. The helium
pressure in the sample bubbler was varied from 500–1020
Torr depending on the desired concentration of Cl2O and
flow rate within the reaction cell.
The flash photolysis apparatus employed has been used
in this laboratory for many studies and is described in detail
elsewhere.5 A brief outline of the features important for these
J. Chem. Phys., Vol. 104, No. 8, 22 February 1996
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.42.202.150 On: Mon, 24 Nov 2014 05:04:06

Nickolaisen et al.: Photolysis of dichlorine monoxide
2859
greater transmission between 350 and 500 nm ͓Fig. 2͑b͔͒. A
solution of CrK͑SO4͒2•12H2O ͑100 g LϪ1͒ had banded trans-
mission between 285–375 and 440–550 nm ͓Fig. 2͑c͔͒.
A. ClO product yield and temporal behavior
Time-dependent concentrations of reactants and products
were measured by long-path uv absorption spectroscopy. The
collimated output of a 150 W short arc xenon lamp was
passed through the reaction cell using White-type optics to
provide a total optical path length of 724 cm. The beam was
passed to a 0.3 m monochromator equipped with a 1024
pixel diode array detector. A 2400 line mmϪ1 diffraction
grating was used which, when coupled with the diode array
detector, resulted in the detection of a spectral band of 30 nm
width. With this detection scheme, the prominent ClO vibra-
tional progression of the ⍀ϭ3/2 subband (A←X) transition
in the region 265–295 nm could be seen in a single scan of
the detector array. By using dedicated memory locations for
data storage, the diode array control electronics and interface
allowed for the acquisition of up to 512 separate scans of the
detector array for each firing of the flashlamp. The temporal
resolution of the experiments was determined by the number
of pixels read in each scan—when the entire 1024 pixels
were read, the minimum temporal resolution was 16.67 ms,
but this value could be reduced to 5 ms when a subset of
only 270 adjacent pixels were used. This lower value re-
sulted from the minimum number of pixels needed in order
to maintain the unambiguous detection of ClO by the obser-
vation of at least five of the vibrational bands centered at 275
nm. Temporal information on the formation and subsequent
removal of the ClO product was obtained by averaging 5 to
100 transient absorption signals following firing of the
flashlamp and fitting the spectra resulting from each scan of
the detector array to a reference spectrum determined from
the known relative cross sections of ClO.13 While absolute
ClO cross sections are dependent on spectrometer resolution,
the measurements here were carried out using the same op-
tical configuration as those employed by Nickolaisen et al. in
which absolute and relative ClO cross sections were mea-
sured with estimated uncertainties of Ϯ20% and Ϯ5%, re-
spectively.
FIG. 2. Flashlamp spectral distributions for different photolysis cell/filter
combinations: ͑a͒ unfiltered quartz cell; ͑b͒ quartz cell with KMnO4 ͑0.2
g LϪ1͒ filter solution; ͑c͒ quartz cell with CrK͑SO4͒2•12H2O ͑100 g LϪ1
filter solution; ͑d͒ unfiltered Pyrex cell.
͒
experiments follows. Identical reaction cells fabricated from
Pyrex or quartz were used in these experiments, depending
on the range of photolysis wavelengths desired. Each cell
consisted of a central reaction region 90.5 cm in length
through which the gas mixture flowed. The reaction cell was
surrounded by three annular jackets. The innermost jacket
comprised an optical filter into which inorganic salt solutions
could be added to provide coarse wavelength filtering of the
spectral distribution of the photolyzing light. The middle
jacket comprised a flashlamp which was used to initiate pho-
todissociation of chlorine oxide. This jacket was filled with
30–50 Torr xenon which was discharged with a 27.5 kV
potential through a 2.4 ␮F capacitor to produce a high inten-
sity broadband flash. The outermost jacket was connected to
a heater/chiller to provide temperature control of the reactor
contents.
The two different cell provided some wavelength control
of the photolysis light—the Pyrex cell cuts off wavelengths
shorter than 300 nm due to the intrinsic properties of the
material, and the quartz cell extended the photolysis cut off
to 200 nm. Spectral distributions of the photolysis flash with
each filter and cell were obtained by recording the spectral
distribution of the photolysis light scattered along the probe
beam detection axis. The associated spectral distributions
from 200 to 400 nm are shown in Figs. 2͑a͒–2͑d͒. To further
limit the spectral distribution of the flashlamp, two salt solu-
tions were used. A solution of KMnO4 ͑0.2 g LϪ1͒ provided a
weak transmission band between 230 and 300 nm with
The ClO product yield as a function of total pressure was
measured under conditions of constant Cl2O concentration.
Total flow of the gas mixture through the cell was adjusted
so that the residence time within the cell was always 5–10 s
shorter than the duty cycle of the flashlamp ensuring that all
photolysis and reaction products are flushed from the cell
between each firing of the flashlamp.
B. Absorption spectra of intermediates
For detection of absorption spectra of possible interme-
diates resulting from Cl2O photolysis, a method was used
that was similar to the one described above with certain
modifications. The monochromator was fitted with a 1200
line mmϪ1 diffraction resulting in a spectral bandwidth of 60
nm. Because of the very small absorption signals involved in
these measurements, it was necessary to perform 100 coad-
J. Chem. Phys., Vol. 104, No. 8, 22 February 1996
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.42.202.150 On: Mon, 24 Nov 2014 05:04:06

2860
Nickolaisen et al.: Photolysis of dichlorine monoxide
FIG. 3. ͑a͒ ClO temporal signals following Cl2O photolysis in unfiltered quartz cell with N2 bath gas at pressures of ͑a͒ 5.0 Torr; ͑b͒ 10.0 Torr; ͑c͒ 25.1 Torr;
͑d͒ 100.1 Torr. ͓Cl2O͔0ϭ5ϫ1014 molecule cmϪ3, Tϭ298 K. Dotted lines indicate the fit of the data to a model which includes a single excited Cl2O† state
undergoing competition between dissociation to products and quenching back to the ground state and the secondary chemistry of the ClOϩClO reactions. All
four temporal signals were fit simultaneously with a single set of rate coefficients. ͑b͒ ClO temporal signals following Cl2O photolysis in the quartz
cell/KMnO4 filter combination with N2 bath gas at pressures of ͑a͒ 5.1 Torr; ͑b͒ 10.1 Torr; ͑c͒ 24.9 Torr; ͑d͒ 99.8 Torr. ͓Cl2O͔0ϭ5ϫ1014 molecule cmϪ3
,
Tϭ298 K. ͑c͒ ClO temporal signals following Cl2O photolysis in the quartz cell/CrK͑SO4͒2•12H2O filter combination with N2 bath gas at pressures of ͑a͒ 5.1
Torr; ͑b͒ 10.1 Torr; ͑c͒ 25.0 Torr; ͑d͒ 100.2 Torr. ͓Cl2O͔0ϭ5ϫ1014 molecule cmϪ3, Tϭ298 K. ͑d͒ ClO temporal signals following Cl2O photolysis in unfiltered
Pyrex cell with N2 bath gas at pressures of ͑a͒ 5.0 Torr; ͑b͒ 10.0 Torr; ͑c͒ 25.1 Torr; ͑d͒ 100.1 Torr. ͓Cl2O͔0ϭ1ϫ1015 molecule cmϪ3, Tϭ253 K.
ditions of the transient signal. To further improve the signal-
to-noise ratio of the spectrum, ten successive scans immedi-
ately preceding the flash were averaged to produce the
baseline I0 spectrum. The transient transmission spectrum,
It , was obtained by averaging the first ten scans immediately
following the flash. An absorption spectrum of the transient
species was derived from the logarithm of the ratio of I0 to
It . By performing the additional averaging, the S/N is sig-
nificantly improved, but the temporal resolution suffers ac-
cordingly. Each scan of the diode array requires 16.67 ms, so
the overall temporal resolution is 167 ms.
individual spectra into a complete absorption spectrum cov-
ering this entire spectral region. Wavelength calibration of
the spectrometer was accomplished by recording the emis-
sion lines from a mercury pen lamp with the diode array for
each monochromator setting. This provided absolute wave-
length markers throughout the entire region of the transient
absorption spectra.
III. RESULTS
A. ClO formation and removal
Because the spectral bandwidth of the monochromator
equipped with a 1200 line mmϪ1 grating was only 60 nm, to
produce a complete absorption spectrum it was necessary to
repeat the acquisition process at a number of monochromator
settings spaced 30 nm apart beginning at 300 nm and extend-
ing to 580 nm. In this manner, the overlap between succes-
sive data sets was sufficient to allow accurate splicing of the
Cl2O photolysis experiments were conducted to deter-
mine the effects of ͑a͒ the spectral distribution of the pho-
tolysis radiation and ͑b͒ the identity and total pressure of
buffer gas on the time-dependent formation and removal of
ClO. Time-dependent ClO concentrations as a function of
total pressure of N2 are shown in Figs. 3͑a͒–3͑d͒ for the four
flashlamp spectral distributions. These signals were recorded
J. Chem. Phys., Vol. 104, No. 8, 22 February 1996
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.42.202.150 On: Mon, 24 Nov 2014 05:04:06

Nickolaisen et al.: Photolysis of dichlorine monoxide
2861
TABLE I. Summary of experimental results as a function of photolysis cell configuration and bath gas identity
and density. ͓ClO͔max is the concentration of ClO at the point of maximum signal, and tmax is the corresponding
reaction time. Yield is defined as ͓ClO͔max/͓Cl2O͔0 .
Cell configuration/
Bath gas
Pressure
͑Torr͒
͓ClO͔
max
tmax
͑s͒
͑1013 molecule cmϪ1
͒
Yield
0.175
0.174
0.166
0.163
0.161
0.127
0.016 5
0.009 32
0.007 66
0.007 20
0.014 7
0.009 84
0.009 76
0.009 40
0.009 20
0.056 7
0.035 0
0.024 4
0.009 71
0.006 97
0.005 53
0.101
Quartz
N2 bath gas
5.0
10.0
15.0
25.1
100.0
5.1
10.1
14.9
25.0
100.1
5.1
10.1
15.3
25.0
100.2
5.0
10.0
15.2
25.1
50.1
100.1
5.0
10.2
15.2
20.3
30.4
50.4
100.3
200.7
400.5
5.1
10.2
15.3
20.6
30.4
50.2
100.3
200.6
400.5
5.3
8.76
8.69
8.29
8.15
8.07
6.36
0.02
0.03
0.02
Ͻ0.01
Ͻ0.01
0.02
0.07
2.3
1.4
Ͻ0.01
0.20
1.20
0.08
0.11
Ͻ0.01
0.02
0.05
0.09
0.10
0.01
Quartz/KMnO4 filter
N2 bath gas
0.824
0.466
0.383
0.360
0.735
0.492
0.488
0.470
0.460
5.67
3.50
2.44
0.971
0.697
0.553
3.42
3.63
2.70
1.02
0.613
0.445
0.209
0.191
0.185
1.75
Quartz/CrK͑SO4͒ filter
2
N2 bath gas
Pyrex/no filter
N2 bath gas
0.02
Pyrex/no filter
O2 bath gas
0.024
0.060
0.084
0.18
0.27
0.31
0.107
0.079 4
0.030 0
0.018 0
0.013 1
0.006 15
0.005 62
0.005 44
0.051 5
0.048 5
0.046 2
0.046 2
0.041 8
0.037 1
0.022 8
0.014 8
0.009 74
0.044 4
0.046 2
0.024 6
0.019 4
0.012 2
0.008 74
0.008 71
0.41
Ͻ0.004
Ͻ0.004
Ͻ0.004
0.032
0.044
0.052
0.096
0.11
0.24
0.57
0.55
0.24
0.21
0.36
0.44
0.51
0.46
0.51
Pyrex
He bath gas
1.65
1.57
1.57
1.42
1.26
0.774
0.504
0.331
1.51
Pyrex
SF6 bath gas
10.1
15.3
20.5
30.6
50.5
100.5
1.57
0.835
0.661
0.416
0.297
0.194
with a time resolution of 5 ms. In all cases, two processes
contribute to the formation of ClO: direct photolysis of Cl2O
and the secondary reaction of atomic chlorine with Cl2O.
The initial rise of the signal due to the ClϩCl2O→ClOϩCl2
reaction is not seen because of the relatively long temporal
resolution of the OMA. Experiments were carried out using
Cl2O concentrations in the range ͑3–10͒ϫ1014
molecule cmϪ3 and pressure range ͑5–400͒ Torr of buffer
gas. Cl2O concentrations were kept at a minimum to avoid
possible effects of Cl2O self-quenching. Most experiments
were carried out at 298 K although a few were conducted at
253 K.
For photolysis in the quartz cell ͓Fig. 3͑a͔͒, the maxi-
mum ClO signal is essentially independent of the total pres-
sure within the reaction cell. The decay of the signal is gov-
erned entirely by the self-reactions of ClO and can be
modeled satisfactorily using the known rate coefficients for
these reactions coupled with the flow of products from the
cell ͑see Sec. IV͒. The time scale for ClO removal decreases
with increasing pressure due to the termolecular component
of the ClO self-reaction. The ClO yield, defined as the maxi-
mum ClO concentration normalized by the initial concentra-
tion of Cl2O, is presented in Table I with other experimental
parameters. For photolysis in the unfiltered quartz cell, the
J. Chem. Phys., Vol. 104, No. 8, 22 February 1996
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.42.202.150 On: Mon, 24 Nov 2014 05:04:06

2862
Nickolaisen et al.: Photolysis of dichlorine monoxide
TABLE II. MRCISD/cc-pVTZ total energies ͑hartree͒ and vertical excita-
tion energies ͑eV͒ for singlet and triplet states of Cl2O, calculated at the
experimental ground-state geometry with CASSCF ͑14 in 9͒ state-averaged
orbitals for pairs of states with the same space and spin symmetry.
State
Etot͑hartree͒
Erel
0.0
State
Etot͑hartree͒
Erel
1 1A1
1 1B1
1 1A2
1 1B2
2 1A1
2 1A2
2 1B1
2 1B2
Ϫ994.365 50
Ϫ994.240 30
Ϫ994.183 42
Ϫ994.182 33
Ϫ994.165 09
Ϫ994.150 88
Ϫ994.096 75
Ϫ994.073 60
1 3B1
1 3A2
1 3B2
1 3A1
2 3A2
2 3A1
2 3B2
2 3B1
Ϫ994.269 04
Ϫ994.217 88
Ϫ994.212 55
Ϫ994.204 31
Ϫ994.158 97
Ϫ994.130 17
Ϫ994.124 75
Ϫ994.120 43
2.625
4.017
4.162
4.386
5.620
6.404
6.552
6.669
3.407
4.954
4.984
5.453
5.840
7.313
7.943
Figure 3͑d͒ shows the ClO temporal signals for photoly-
sis in the Pyrex cell. The behavior is similar to that observed
for the filtered quartz cell data, namely, that the yield de-
creases and the formation time to the maximum ClO signal
increases as the bath gas density increases ͑see Table I͒. Use
of the Pyrex cell limits the photolysis wavelengths to 300 nm
thereby accessing only the ‘‘weak’’ transition on the shoulder
of the 254 nm band and the bands at 410 and 530 nm. A
series of measurements were also made using the Pyrex cell
in which the temporal ClO behavior was monitored using
either He, O2, or SF6 as the bath gas. The results are sum-
marized in Table I. In each case, the yield decreased and the
observed product formation time increased with increasing
bath gas pressure. The effect of increased pressure on the
experimental observables of yield and product formation
time was more gradual with He as the bath gas, and was
most pronounced with SF6. For both O2 and SF6, the yield
actually increased when the pressure was varied from 5 to 10
Torr, but then dropped rapidly at higher pressures.
FIG. 4. ͑a͒ Total absorption spectrum collected from the first ten scans of
the diode array immediately following firing of the flashlamp.
͓Cl2O͔0ϭ1ϫ1015 molecule cmϪ3, cell temperature was 253 K, and total
pressure was 5.0 Torr with N2 as the bath gas. ͑b͒ OClO absorption spectrum
collected approximately 7 s following firing of the flashlamp when ͓OClO͔
is near its maximum. OClO is a product of the ClOϩClO reaction. ͑c͒ Cl2O
metastable intermediate absorption spectrum determined by subtracting an
appropriately scaled OClO spectrum ͓Fig. 8͑b͔͒ from the total absorption
spectrum ͓Fig. 8͑a͔͒.
Experiments were conducted to determine if the
pressure-dependent ClO yields and temporal variations ob-
served in the Pyrex cell were due to experimental artifacts
peculiar to the cell material. A Pyrex sleeve was placed in-
side the quartz cell and the experiments were repeated. The
results were identical with those obtained in the Pyrex cell
ruling out an artifacts associated with the cell itself ͑other
than the cell material͒. Another test was made in which an
opaque liner was placed inside the Pyrex cell to block the
photolysis light from entering. The observed formation of
ClO was negligible indicating that the dissociation of Cl2O
occurred by photolysis rather than some other process such
as some sort of electrostatic discharge arising from the high
voltage pulse on the xenon flash lamp.
yield is essentially constant over the density range
͑1–30͒ϫ1017 molecule cmϪ3, and the time to the maximum
signal becomes shorter with increasing pressure.
Figures 3͑b͒ and 3͑c͒ show ClO data collected using the
quartz cell with the KMnO4 and CrK͑SO4͒2•12H2O filter so-
lutions, respectively. Yield data are given in Table I. There is
a marked difference when compared to the unfiltered quartz
data of Fig. 3͑a͒. For each experimental configuration, the
data as a function of pressure display two prominent features.
First, the ClO yield, as defined above, decreases with in-
creasing bath gas density, and second, the time required to
reach the maximum ClO signal increases as the bath gas
density increases. Photolysis in the unfiltered quartz cell ac-
cesses both the strong absorption band of Cl2O at 254 nm as
well as the shoulder of the strong absorption feature in the
region 200 to 225 nm ͑see Fig. 1͒. Photolysis using the filter
solutions completely removes any contribution from the 200
to 225 nm band. The KMnO4 filter severely limits the con-
tribution from the 254 nm band, and the CrK͑SO4͒2 filter
eliminates this contribution altogether. The difference in the
product yield behavior between the unfiltered quartz cell and
these filtered configurations can be related to the differences
in the regions of the absorption spectrum accessed by the
photolysis radiation.
B. Metastable intermediate formation
The dependence of the ClO yield on the buffer gas den-
sity suggests that in addition to a direct dissociation pathway,
a metastable intermediate is formed which subsequently de-
composes to ClO or quenches to one or more other states.
Using the OMA, a spectral search was carried out for tran-
sient absorbers to identify other species or metastable states
which might be formed from the photolysis. Based on simu-
lations of the secondary chemistry of the species expected to
J. Chem. Phys., Vol. 104, No. 8, 22 February 1996
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.42.202.150 On: Mon, 24 Nov 2014 05:04:06

Nickolaisen et al.: Photolysis of dichlorine monoxide
TABLE III. MRCISD/cc-pVTZ vertical excitation energies ͑eV͒ and oscil-
2863
IV. DISCUSSION
lator strengths for singlet states ͑relative to S0͒ and triplet states ͑relative to
a
T0͒ calculated at the experimental ground-state geometry for Cl2O with
The interpretation of the results of this work is heavily
reliant upon an understanding of the energetics and symme-
try properties of the Cl2O excited states. We begin by dis-
cussing the methods and results of the ab initio calculations,
then use these results to interpret the ClO formation kinetics
and the spectrum of the observed metastable intermediate.
three-state CASSCF ͑14 in 9͒ averaged orbitals.
State
Erel
0.0
f
State
Erel
0.0
f
1 1A1
1 1B1
1 1A2
1 1B2
2 1A1
2 1A2
2 1B1
2 1B2
1 3B1
1 3A2
1 3B2
1 3A1
2 3A2
2 3A1
2 3B2
2 3B1
a
3.419
4.969
4.983
5.453
5.884
7.345
7.981
1.42ϫ10Ϫ4
0.0
1.417
1.559
1.775
3.009
3.819
3.962
4.044
8.81ϫ10Ϫ4
0.0
7.75ϫ10Ϫ3
8.5ϫ10Ϫ5
0.0
3.3ϫ10Ϫ6
1.03ϫ10Ϫ1
8.0ϫ10Ϫ5
0.0
A. Computational methods
Theoretical calculations for excited states of Cl2O were
carried out using MOLPRO,14 a set of ab initio electronic
structure programs. Complete-active-space self-consistent-
5.26ϫ10Ϫ3
6.0ϫ10Ϫ6
5.37ϫ10Ϫ3
15
field ͑CASSCF͒ calculations were performed for the two
aT0(1 3B1) state is 2.625 eV above S0(1 1A1).
states of lowest energy within each irreducible representation
of the C2v point group, for both the singlet and triplet mani-
folds, using the experimentally determined ground-state mo-
lecular geometry. Subsequently the CASSCF molecular or-
bitals ͑MOs͒ were used in multireference configuration-
interaction ͑MRCI͒ calculations for these states.16 These
were of the internally contracted type, and used the projec-
tion method of Knowles and Werner.17
be present in the cell following the flash, the only species
with absorptions in the region 350 to 550 nm are Cl2O itself,
and OClO and Cl2 which are products of the ClO self-
reaction. The absorption spectrum of ground state Cl2O is
broad with no sharp distinguishing features and will result
only in gradual variations of the baseline. Similarly, the Cl2
spectrum is a broad continuum and will produce no sharp
features in the absorption spectrum. OClO has a structured
spectrum from 270 to 475 nm corresponding to a vibrational
progression in the A←X band. To remove the OClO contri-
bution to the product spectrum, a reference spectrum was
produced from the OMA data collected at times greater than
7.5 s following the flash when the reactions which result in
OClO production were essentially complete. The OClO ref-
erence spectrum was appropriately scaled and subtracted
from the total absorption spectrum. Figures 4͑a͒–4͑c͒ dis-
plays three spectra: ͑a͒ the total absorption spectrum for all
species immediately following the flash, ͑b͒ the OClO refer-
ence spectrum, and ͑c͒ the spectrum resulting from the re-
moval of the OClO contribution. Figure 4͑c͒ shows a weak
absorption spectrum with a complex series of vibrational
progressions over the spectral range 400–600 nm. The tem-
poral dependence of the residual spectrum was measured and
found to persist for about 500 ms. The absorption spectrum
of the metastable maximized at low pressure ͑5 Torr N2 at a
temperature of 253 K͒.
The spectrum of Fig. 4͑c͒ cannot be assigned to any of
the other species that might be produced from secondary
reactions of Cl2O photolysis products ͑ClOO, ClO3, Cl2O2,
etc.͒ or from minor impurities in the Cl2O. The structured
bands are well above the noise level of the data and were
reproduced a number of times using Cl2O samples prepared
at different times. In addition, all of the absorption bands in
the new spectrum showed the same time dependence indicat-
ing the presence of a single absorber. The increase in base-
line absorption at shorter wavelengths is attributed to
changes in the concentration of ground state Cl2O; the Cl2O
spectrum was not subtracted out because the broad absorp-
tion bands of ground state Cl2O do not interfere with the
structured spectrum of the intermediate.
1
The Hartree–Fock electronic configuration for the X A1
2
2
2
2
ground state of Cl2O is ͑core͒͑6a1͒ ͑5b2͒ ͑7a1͒ ͑2b1͒
2
2
2
2
2
2
0
͑6b2͒ ͑8a1͒ ͑2a2͒ ͑9a1͒ ͑7b2͒ ͑3b1͒ ͑8b2͒ ͑10a1͒0, where the
9a1 and 6b2 MOs are ClO ␴ bonding and the unoccupied
8b2 and 10a1 MOs are ClO ␴ antibonding. The active space
for full-valence CASSCF calculations would comprise all
these MOs, giving 20 electrons in 12 orbitals. Inspection of
the CASSCF CI vectors for this active space showed that, for
all the states calculated, the MOs 6a1 , 7a1 , and 5b2 , ͑aris-
ing from the Cl 3s and O 2s orbitals͒ remained doubly oc-
cupied in all significant configurations. By excluding these
three MOs the active space was reduced to 14 electrons in 9
orbitals, arising from the valence p subshell of each atom. A
total of between 158 and 167 configuration state functions
͑CSFs͒ were generated from this active space for the various
states calculated. Further examination of the CASSCF CI
vectors showed that for all symmetry species the predomi-
nant configurations had at least two electrons in the orbitals
8a1 and 6b2 . The final reference space adopted for the
MRCI calculations therefore comprised all the CASSCF
CSFs except those in which the combined occupancy of the
8a1 and 6b2 was less than two; elimination of these configu-
rations had only a minor effect on the total number of CSFs.
The calculations were performed using the Dunning
correlation-consistent polarized valence triple-zeta ͑cc-
pVTZ͒ basis,18 which has the form (10s5p2d1f)/
[4s3p2d1f] for oxygen and (15s9p2d1f)/[5s4p2d1f] for
chlorine; this gave a total number of 98 basis functions. To
check for basis-set effects in the results, calculations were
also made with a smaller valence double-zeta basis ͑involv-
ing only 50 basis functions͒ but the results were found to
differ very little.
B. Computational results and discussion
Table II shows total energies and vertical transition en-
ergies, calculated at the MRCISD/cc-pVTZ level, for the fol-
J. Chem. Phys., Vol. 104, No. 8, 22 February 1996
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.42.202.150 On: Mon, 24 Nov 2014 05:04:06

2864
Nickolaisen et al.: Photolysis of dichlorine monoxide
lowing excited states: 1 1A1 , 2 1A1 , 1 1B1 , 2 1B1 , 1 1B2 ,
2 1B2 , 1 1A2 , 2 1A2 , 1 3A1 , 2 3A1 , 1 3B1 , 2 3B1 , 1 3B2 ,
1 3B2 , 1 3A2 , and 2 3A2 . State-averaged orbitals were used
for each of the different space and spin symmetries in
order to avoid bias towards the lower energy states. First,
eight separate two-state CASSCF ͑14 in 9͒ calculations
7.345 eV for the 2 1B1 state. The peak observed at 4.88 eV is
consistent with the 1 1B2 state calculated to be at 4.983 eV.
Examination of the computed oscillator strengths indicates
that these two states should indeed have the strongest absorp-
tions in agreement with the observed spectrum. However, the
ordering is reversed—ab initio results predict the 1 1B2 state
to have a slightly larger absorption cross section relative to
the 2 1B1 state whereas the experimental spectrum indicates
a larger oscillator strength for the higher energy state.
The computed oscillator strengths indicate that the next
strongest absorption following the two large features is the
transition to the 1 1B1 state. The calculated energy of this
transition is 3.419 eV which does not directly correlate with
any features of the observed absorption spectrum. However,
the shoulder to the low energy side of the 254 nm peak is of
sufficient strength that this feature most likely arises from the
1 1B1←1 1A1 transition. If this is indeed the case, then the
calculated transition energy of this state is low by several
tenths of an eV. The transition to the 2 1A1 state at 5.453 eV
also has sufficient oscillator strength to be observed experi-
mentally, but this absorption is buried in the shoulder of the
large feature at 7.24 eV.
1
were performed for (X A1ϩ2 1A1), (1 1B1ϩ2 1B1),
(1 1B2ϩ2 1B2), (1 1A2ϩ2 1A2), (1 3A1ϩ2 3A1), (1 3B1
ϩ2 3B1), (1 3B2ϩ2 3B2), and (1 3A2ϩ2 3A2). The MOs
from each of these calculations were then used in MRCISD
calculations for the two states; e.g., the MOs from the
(1 1B2ϩ2 1B2) CASSCF calculation were used to calculate
the energies of the first and second 1B2 states at the
MRCISD level.
Transition dipole moments ͑and hence oscillator
strengths͒ between the lowest-energy state and each higher
state, within both the singlet and triplet manifolds, were cal-
culated at the MRCISD/cc-pVTZ level, as shown in Table
III. It is necessary to employ the same orbitals for both states
involved in each transition.19 Thus to obtain orbitals aver-
aged over both the lower and upper state required a
further three-state CASSCF ͑14 in 9͒ calculation, i.e.,
1
1
1
(X A1ϩ1 1B1ϩ2 1B1), (X A1ϩ1 1B2ϩ2 1B2), (X A1
ϩ1 1A2ϩ2 1A2), (1 3B1ϩ1 3A1ϩ2 3A1), (1 3B1ϩ1 3B2
ϩ2 3B2), and (1 3B1ϩ1 3A2ϩ2 3A2), unless the upper and
lower states were already of the same symmetry. To calculate
the transition dipole moment for the transition
(2 3A2←1 3B1), for example, an MRCISD wave function for
each state was determined using the orbitals from the
(1 3B1ϩ1 3A2ϩ2 3A2) calculation. Since the three-state
CASSCF calculations give orbitals that are less well adapted
to each individual state than are those from the two-state
CASSCF calculations, the resulting energies are slightly less
satisfactory for each state. However, comparison of the tran-
sition energies in Tables II and III reveals no significant dif-
ferences. Moreover, it was found that very similar results
were obtained for transition energies and oscillator strengths
at the CASSCF level using either the ͑14 in 9͒ or the larger
͑20 in 12͒ active space, and with either the cc-pVTZ or the
smaller cc-pVDZ basis; these findings ͑not shown͒ confirm
the essential reliability of the results presented.
The weak absorption peaks at 2.3 and 2.9 eV cannot be
assigned to singlet←singlet transitions because the 1 1B1
state is the lowest excited state in the singlet manifold, and
the energy of this state is substantially higher than the ob-
served transition energies of these two features. The 2.9 eV
peak corresponds to a spin forbidden transition to the lowest
lying triplet state, 1 3B1 . The oscillator strength for this
triplet←singlet transition was not determined in this study,
but this assignment is consistent with the calculated transi-
tion energy of 2.625 eV and the small absorption cross sec-
tion of 1ϫ10Ϫ20 cm2 expected for a forbidden transition such
as this. However, that excitation of this triplet state occurs is
an indication that the 1 3B1 state is not strictly triplet in
nature, but rather has a mixed configuration containing some
singlet character. The very weak feature at 2.3 eV cannot be
assigned to a transition from the ground electronic state since
the excitation energy is significantly lower than any of the
calculated excited state energies in either the singlet or triplet
manifolds. This absorption may arise from a transition be-
tween two excited electronic states where the lower lying
state is prepared by excitation from the ground state.
The lowest energy triplet state calculated was 1 3B1 . The
predominant configuration in the MRCISD vector for this
state corresponds to an excitation from the (3b1)2 nonbond-
ing orbital of the Hartree–Fock ground state to the
C. ClO formation kinetics
0
*
(10a1) ␴ ͑ClO͒ antibonding orbital. The transition with the
largest oscillator strength is 2 3A2←1 3B1 ; this transition
corresponds to excitation of an electron from the doubly oc-
cupied (2a2)2 orbital in the 1 3B1 state to the singly occu-
pied (3b1)1 orbital.
Our experiments focused on the production of ClO from
the broadband pulsed photolysis of Cl2O using different pho-
tolytic spectral distributions extending from the visible to
200 nm. The ClO time dependence was dominated by two
major effects: The kinetics of ClO formation which, when
photolyzed at longer wavelengths, were determined prima-
rily by the excited state dynamics of Cl2O and the kinetics of
ClO removal which were dominated by ClO self-reaction.
These processes will be discussed below.
The computed vertical excitation energies and oscillator
strengths of Table III may be used to interpret the observed
Cl2O absorption spectrum shown in Fig. 1. The experimental
spectrum has five prominent features in the region of interest
to this study—absorption peaks at 171 nm ͑7.24 eV͒, 254 nm
͑4.88 eV͒, 420 nm ͑2.9 eV͒, and 540 nm ͑2.3 eV͒, and a
shoulder on the 254 nm peak at 300 nm ͑4.1 eV͒. The largest
peak at 7.24 eV correlates well with the calculated energy of
The time dependence of ClO production was influenced
strongly by the spectral domain of the photolysis radiation,
the bath gas pressure and the bath gas identity. Restricting
J. Chem. Phys., Vol. 104, No. 8, 22 February 1996
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.42.202.150 On: Mon, 24 Nov 2014 05:04:06

Nickolaisen et al.: Photolysis of dichlorine monoxide
2865
TABLE IV. Mechanism for the simulation of ClO kinetic decay in the
photolysis of Cl2O.
the spectrum of the photolysis radiation to wavelengths
longer than about 250 nm results in significant changes in the
production of ClO. First, the time scale for the formation of
Reaction
k ͑cm3 moleculeϪ1 sϪ1
͒
Reference
2
ClO (X ⌸) changes from instantaneous on the scale of the
Cl2Oϩh␯→ClOϩCl
Cl2Oϩh␯→Cl2ϩO
⌽ϭ0.75
⌽ϭ0.25
5
5
OMA temporal resolution to values in the range of 50–200
ms. The time at which the ClO maximum is reached in-
creases with increasing bath gas density until about 100 Torr
at which point the time to reach the maximum decreased.
Second, the maximum photolytic yield of ClO changes from
being independent of the bath gas density to inversely depen-
dent on the density. These observations strongly suggest that
Cl2O photolysis beyond 250 nm results in the formation of
metastable states which undergo competitive collisional de-
activation and unimolecular decomposition to form ClO and
possibly other products. Due to the broadband nature of the
photolysis spectrum, the wavelength dependence of the ex-
citation leading to the Cl2O metastable cannot be determined.
The observed behavior is a convolution of the photolytic
intensity distribution which increases rapidly at longer wave-
lengths, the absorption cross sections which decrease rapidly
at longer wavelengths, and the energy dependence of the
excitation leading to the curve crossing which is unknown.
Insight into the excited state dynamics arising from the
long-wavelength excitation may be obtained from a kinetic
simulation of the ClO temporal profile and its dependence on
the buffer gas density. The simplest mechanism for the for-
mation of ClO product consists of a single metastable excited
state with competition between quenching and dissociation,
i.e.,
Cl2OϩCl→ClOϩCl2
Cl2OϩO→2ClO
9.8ϫ10Ϫ11
3.5ϫ10Ϫ12
4.9ϫ10Ϫ15
8.0ϫ10Ϫ15
3.5ϫ10Ϫ15
͑0.99Ϫ3.15͒ϫ10Ϫ32a
3.8ϫ10Ϫ11
1.0ϫ10Ϫ13
1.2ϫ10Ϫ11
2.3ϫ10Ϫ10
1.0ϫ10Ϫ10
1.1ϫ10Ϫ12
2.3ϫ10Ϫ18
2.7ϫ10Ϫ33a
1.9ϫ10Ϫ31a
20
20
13
13
13
13
20
20
20
20
20
20
20
20
21
ClOϩClO→Cl2ϩO2
ClOϩClO→ClOOϩCl
ClOϩClO→OClOϩCl
ClOϩClOϩM→Cl2O2ϩM
OϩClO→ClϩO2
OϩOClO→ClOϩO2
ClϩClOO→2ClO
ClϩClOO→Cl2ϩO2
ClϩCl2O2→ClOOϩCl2
ClOOϩM→ClϩO2ϩM
Cl2O2ϩM→2ClOϩM
ClϩO2ϩM→ClOOϩM
OϩOClOϩM→ClO3ϩM
aUnits are cm6 moleculeϪ2 sϪ1
.
processes are given in Table IV. The simulations were per-
formed by simultaneously fitting the ClO time dependent
signals at pressures of 5, 10, 25, and 100 Torr to this mecha-
nism using a nonlinear least-squares technique. The param-
eters varied in the fitting routine were the initial amount of
͓Cl2O†͔0, kd , and kq . The resulting values were
͓Cl2O†͔0ϭ5.4ϫ1013 molecule cmϪ3, kdϭ1.5ϫ1013 sϪ1, and
kqϭ4.4ϫ10Ϫ14 cm3 moleculeϪ1 sϪ1. The values for kd and
kq are reasonable, but may not accurately represent the rate
coefficients for these physical processes because the hyper-
surface used in the fitting routine is not tightly constrained
by the data.
While this mechanism predicts the ClO yield and tem-
poral behavior for photolysis at short wavelengths, it does
not describe the experimental data for the other photolysis
configurations in which absorption to the 2 1B1 state is not
involved. To model the temporal behavior correctly, it was
necessary to include two excited states in the mechanism.
The first state is populated either directly by photolysis or by
intersystem crossing from the initially prepared singlet upper
state. The second state is produced by collisional relaxation
from the first state and may dissociate to products or undergo
collisional quenching to the ground state. Examination of the
energy level diagram in Fig. 5 suggests that plausible candi-
dates for the two metastable excited states are the 1 3A2 and
1 3B1 states, respectively, and that the initially prepared sin-
glet state in this wavelength region is 1 1B1 . The complete
mechanism is
Cl2Oϩh␯→Cl2O† excitation,
k
d
Cl2O†→ClOϩCl dissociation,
k
q
Cl2O†ϩM→Cl2OϩM quenching.
The time dependence for ClO formation is given by
kd Cl O†
͓
͔
2
0
d
q
ClO ϭ
1ϪeϪ͑k ϩk ͓M͔͒t͒.
͑
͓
͔
kdϩkq
M
͓
͔
The kinetics of the ClO removal are dominated by the self-
reaction of ClO. As discussed by Nickolaisen et al.,13 the
detailed mechanism of the ClO self-reaction consists of both
bimolecular and termolecular components at temperatures
above about 250 K. The detailed reaction mechanism is
given in Table IV. The time scale for ClO removal by ClO
self-reaction is sufficiently long ͑tϾ0.1 s͒ that the ClO re-
moval mechanism can be decoupled from the excited state
dynamics which produce ClO at long wavelengths. This
mechanism for ClO formation coupled with the ClO self-
reaction kinetics predicts that as the pressure is increased, the
time to reach the maximum ClO signal will decrease. The
observed ClO signals for photolysis in the unfiltered quartz
cell conform to the time dependence predicted by this equa-
tion. Figure 3͑a͒ illustrates the fit of the unfiltered quartz cell
data to a mechanism that included the above elementary
steps and all chemical processes involved in the ClO self-
reaction. Reactions and rate coefficients for the ClO removal
Cl O 1 1A ͒ϩh␯→Cl O 1 1B ͒,
͑
͑
2
1
2
1
Cl O 1 1B ͒→Cl O 1 3A ͒,
͑
͑
2
1
2
2
Cl O 1 3A ͒→ClOϩCl,
͑
2
2
Cl O 1 3A ͒ϩM→Cl O 1 3B ͒ϩM,
͑
͑
2
2
2
1
Cl O 1 3B ͒→ClOϩCl,
͑
2
1
Cl O 1 3B ϩM→Cl O 1 1A ϩM.
͒
͒
1
͑
͑
2
1
2
J. Chem. Phys., Vol. 104, No. 8, 22 February 1996
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.42.202.150 On: Mon, 24 Nov 2014 05:04:06

2866
Nickolaisen et al.: Photolysis of dichlorine monoxide
FIG. 5. Energy diagram of Cl2O electronic states. Vertical excitation ener-
gies are given in eV, and oscillator strengths are included in parentheses for
each state. Bold lines indicate processes involved in the long wavelength
FIG. 6. ClO temporal signals resulting from model simulations of the reac-
tion mechanism which includes quenching to a Cl2O metastable intermedi-
ate from the initially excited electronic state, dissociation of the intermediate
state, and quenching of the intermediate state to the ground state. Upper
panel is the experimental data of Figure 3͑d͒ with an expanded time axis.
Lower panel are the simulated data. Total pressures are ͑a͒ 5 Torr; ͑b͒ 10
Torr; ͑c͒ 25 Torr; ͑d͒ 100 Torr.
photolysis. h␯
represents excitation by the xenon flashlamp, M indi-
excitation
cates collisional quenching, ISC indicates intersystem crossing, and h␯
probe
represents absorption to produce the observed metastable spectrum.
Figure 5 schematically indicates the process occurring during
Cl2O photolysis for wavelengths greater than 300 nm. Initial
excitation from the ground state is dominated by transition to
the 1 1B1 state with a small fraction ͑from the relative ab-
sorption cross sections, perhaps 1%–2%͒ of population in
the 1 3B1 state. As mentioned previously, the actual energy
of the 1 1B1 state is higher than shown in Fig. 5, and is
probably much closer to the wavelength cut off of the Pyrex
cell at 300 nm. Thus the 1 1B1 and 1 3A2 states will be in
near resonance so that intersystem crossing to the triplet
manifold from the initially prepared singlet state occurs. This
process is very efficient because of the energy overlap of
rovibrational levels within the 1 1B1 and 1 3A2 states, the
density of rovibrational states, and the probable mixed char-
acter of the 1 3A2 state. As a result, the 1 1B1 state undergoes
competition between intersystem crossing to the 1 3A2 state
and collisional quenching to the ground state.
shown in Fig. 6. The qualitative features of the experimental
data are reproduced by this mechanism, namely, that the ClO
yield decreases and the time required to attain the maximum
ClO signal initially increases and then decreases with in-
creasing bath gas density. The model could not be made to fit
the data quantitatively. This is due to the simplified nature of
the excited state mechanism and the computational difficul-
ties involved in deriving a best fit of four data sets to a
six-parameter mechanism.
The transient absorption spectrum shown in Fig. 4͑c͒ is
important in the interpretation of the dissociation mecha-
nism. As mentioned previously, the transient spectrum is
highly structured and cannot be attributed to any of the
known transient radical or stable molecular species produced
chemically in the above mechanism. A possible explanation
is that the transient spectrum arises from the formation of a
ClO•Cl2O complex since, immediately following the pho-
tolysis pulse, the reaction cell contains relatively high con-
centrations of both ClO and Cl2O. If adduct formation was
the source of the observed transient spectrum, the adduct
would also be seen in the quartz cell in even higher concen-
trations because more ClO is produced under these photoly-
sis conditions. However, careful examination of the data fol-
lowing quartz cell photolysis provided no evidence of the
transient spectrum seen in the Pyrex cell. Adduct formation
also fails to describe the observed increase in the ClO prod-
uct formation rate with increasing pressure. A ClO•Cl2O
complex would have the same kinetic behavior as a function
of pressure as ClO dimer formation. Simulations of a mecha-
nism that includes adduct formation indicate that the oppo-
site effect on the product formation rate should be observed,
namely, that the rate of product formation becomes faster
Once in the triplet manifold, Cl2O(1 3A2) either dissoci-
ates to products or is collisionally quenched to the lowest
lying 1 3B1 state. The 1 3B1 state, in turn, may also dissoci-
ate to products or be quenched to the ground 1 1A1 state. The
quenching step is quite inefficient because of the spin forbid-
den nature of the transition and the large energy difference
between states, so that this process becomes significant only
at higher pressures. We know that the lifetime of the 1 3B1
state is quite long because it is this state that is the source of
the intermediate absorption spectrum measured in these ex-
periments. Hence, the dissociation process from this lowest
triplet state must also be inefficient. This mechanism was
used to simulate the ClO temporal signals of Fig. 3͑d͒ for
photolysis in the Pyrex cell at 253 K with nitrogen as the
bath gas over the pressure range 5–100 Torr. Results are
J. Chem. Phys., Vol. 104, No. 8, 22 February 1996
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.42.202.150 On: Mon, 24 Nov 2014 05:04:06

Nickolaisen et al.: Photolysis of dichlorine monoxide
2867
283 cmϪ1 mode. This frequency is consistent with a ClOCl
bending motion and indicates a significant change in the
bond angle between the upper and lower electronic states.
Another interesting aspect of the spectrum in this region is
the presence of a strong progression at 886 cmϪ1, which may
indicate that 443 cmϪ1 corresponds to an antisymmetric ClO
stretching motion. A more quantitative analysis of the spec-
trum requires higher resolution data.
The ab initio excitation energies and oscillator strengths
given in Tables II and III for electronic transitions within the
Cl2O triplet manifold are consistent with the experimentally
observed transient absorption spectrum. There are only two
excited electronic states energetically accessible after Cl2O
photoexcitation with wavelengths longer than 350 nm: 1 1B1
͑S1 ; ␭threshϭ364 nm͒ and 1 3B1 ͑T0 ; ␭threshϭ472 nm͒. While
we cannot eliminate singlet–singlet transitions originating
from S1 as the source of the metastable spectrum, the
2 3A2←1 3B1 transition has an unusually large oscillator
strength and is at least an order of magnitude stronger than
any other singlet or triplet transition listed. Additionally, the
calculated transition energy, ϳ3.0 eV ͑24 200 cmϪ1͒, lies
within the range of the experimentally observed transient
spectrum. Our calculations are still exploratory at this stage
and should not be used for definitive spectral assignments,
but they do suggest that the upper state in the 22 000–26 000
cmϪ1 transition may possess a large fraction of 2 3A2 char-
acter.
The ab initio calculations also suggest an explanation for
the complicated, low intensity features present below 22 000
cmϪ1. Three excited triplet configurations were identified at
energies below the 2 3A2 state. Any or all of these states
could mix with the 2 3A2 state and borrow oscillator
strength, thereby producing observable absorptions and com-
plicated vibronic patterns. Similarly, mixing between the
nearly degenerate 2 1A2 and 2 3A2 states might provide suf-
ficient oscillator strength to produce some features in the
otherwise forbidden 2 1A2←1 3B1 transition. Severe pertur-
bations in the vibrational spacings of the spectrum might also
result from mixing of the metastable eigenstates with the
manifold of dark S0 and/or T0 levels. Further experimental
and theoretical studies are required to uncover these aspects
of Cl2O photochemistry.
The observed pressure dependent photolysis yields for
Cl2O may have important implications for stratospheric
chemistry. For most temporary reservoirs of odd hydrogen,
nitrogen, chlorine and bromine in the lower stratosphere,
photolysis is the dominant loss mechanism. Below about 26
km, photolysis is restricted to wavelengths longer than about
290 nm due to absorption by ozone at higher altitudes.
Above this wavelength, absorption cross sections are gener-
ally less than 5ϫ10Ϫ20 cm2 moleculeϪ1, decreasing rapidly
with wavelength. Excitations in this region may be domi-
nated by weak transitions whose dissociation dynamics are
highly uncertain because quantum yield studies typically ad-
dress the more intense region of the continuum. Species that
fall into this category include ClONO2, N2O5, H2O2, HNO3,
and HO2NO2. Using the same technique employed for the
Cl2O studies, we have observed that the yields of ClO and
FIG. 7. Vibronic band assignments are shown for the dominant progressions
of the metastable Cl2O spectrum observed between 22 000 and 26 000
cmϪ1. The long progressions A and B have frequencies characteristic of a
Cl–O stretching mode and suggest significant differences between the T0
and T1 equilibrium structures.
with increased pressure. These two points argue against a
ClO•Cl2O complex as the source of the transient spectrum.
D. Spectral analysis
The spectrum of the transient absorber shown in Fig.
4͑c͒ displays extensive, complicated vibronic structure.
Closer inspection shows that the spectrum separates into
three distinguishable sections. The lowest energy region
͑16 600–20 000 cmϪ1͒ is characterized by a series of low
intensity features with ϳ100 and ϳ300 cmϪ1 spacing, but
lacking a well-defined origin. A positive identification of any
regular progressions is complicated by the near resonance of
3ϫtimes 100 with 300 cmϪ1 and the fact that these bands
overlap one another in most instances.
The intermediate energy region ͑20 000–22 000 cmϪ1
͒
has the type of overall intensity contour one would expect
for a Franck–Condon envelope. The strength of individual
features increases from a minimum near 20 000 cmϪ1, peaks
near 21 000 cmϪ1, then slowly decreases with increasing en-
ergy. The number of features in this energy range, however,
is far larger than one might expect for the spectrum of a
well-behaved triatomic molecule. We have identified frag-
mentary series having 110, 325, and 768 cmϪ1 spacings in
this region as well as a number of series having 345–350
cmϪ1 spacings but we have been unable to establish a unique
origin for the bands in this region.
The spectrum changes significantly in the 22 000–
26 000 cmϪ1 region. An expanded view of this region is
shown in Fig. 7. The first few features in this region are quite
intense and well separated from their neighbors. It was pos-
sible to identify progressions having 283, 443, and 505 cmϪ1
differences using the strong feature at 22 322 cmϪ1 as a start-
ing point. Incorporating these three frequencies into an har-
monic oscillator calculation, 31 features were fitted within
the experimental resolution of the spectrum. The dominant
progression was one with up to 8 quanta of excitation in the
J. Chem. Phys., Vol. 104, No. 8, 22 February 1996
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.42.202.150 On: Mon, 24 Nov 2014 05:04:06

2868
Nickolaisen et al.: Photolysis of dichlorine monoxide
NO3 from the photolysis of ClONO2 at wavelengths longer
than 300 nm are also pressure dependent.22 In order to assess
the importance of this mechanism for stratospheric chemistry
it is necessary to carry out quantum yield studies using
monochromatic excitation over a range of wavelengths and
bath gas densities appropriate for the lower stratosphere.
7 A. I. Chichinin, Chem. Phys. Lett. 209, 459 ͑1993͒.
8 C.-L. Lin, J. Chem. Eng. Data 21, 411 ͑1976͒.
9 L. T. Molina and M. J. Molina, J. Phys. Chem. 82, 2410 ͑1978͒.
10 H. D. Knauth, H. Alberti, and H. Clausen, J. Phys. Chem. 83, 1604 ͑1979͒.
11 J. B. Nee, J. Quant. Spectrosc. Radiat. Transfer 46, 55 ͑1991͒.
12 G. H. Cady Inorg. Synth. 5, 156 ͑1957͒.
13 S. L. Nickolaisen, R. R. Friedl, and S. P. Sander, J. Phys. Chem. 98, 155
͑1994͒.
14
MOLPRO: a package of ab initio programs written by H.-J. Werner and P. J.
ACKNOWLEDGMENTS
Knowles, with contributions of J. Almlof, R. Amos, S. Elbert, K. Hampel,
W. Meyer, K. Peterson, R. Pitzer, and A. Stone.
We wish to thank the JPL kinetics and photochemistry
group, C. Wittig and H. Reisler for helpful discussions and
R. Beaudet for his preliminary analysis of the metastable
spectrum. Part of the research described in this paper was
carried out at the Jet Propulsion Laboratory, California Insti-
tute of Technology, under contract to the National Aeronau-
tics and Space Administration.
15 H.-J. Werner and P. J. Knowles, J. Chem. Phys. 82, 5053 ͑1985͒; P. J.
Knowles and H.-J. Werner, Chem. Phys. Lett. 115, 259 ͑1985͒.
16 H.-J. Werner and P. J. Knowles, J. Chem. Phys. 89, 5803 ͑1988͒; P. J.
Knowles and H.-J. Werner, Chem. Phys. Lett. 145, 514 ͑1988͒.
17 P. J. Knowles and H. J. Werner, Theor. Chim. Acta 84, 95 ͑1988͒.
18 T. H. Dunning, J. Chem. Phys. 90, 1007 ͑1989͒; D. E. Woon and T. H.
Dunning ibid. 98, 1358 ͑1993͒.
19 C. W. Bauschlicher, Jr. and S. R. Langhoff, Chem. Rev. 91, 701 ͑1991͒.
20 W. B. DeMore, S. P. Sander, D. M. Golden, R. F. Hampson, M. J. Kurylo,
C. J. Howard, A. R. Ravishankara, C. E. Kolb, and M. J. Molina, Chemi-
cal Kinetics and Photochemical Data for Use in Stratospheric Modeling—
Evaluation Number 11, JPL Publication 94-26, Jet Propulsion Laboratory,
Pasadena ͑1994͒.
1 W. Finkelnburg, H. J. Schumacher, and G. Stieger, Z. Phys. Chem. B 15,
127 ͑1931͒.
2 H. J. Schumacher and R. V. Townend, Z. Phys. Chem. B 20, 375 ͑1933͒.
3 F. H. C. Edgecomb, R. G. W. Norrish, and B. A. Thrush, Proc. R. Soc.
London Ser. A 243, 24 ͑1957͒.
21 A. J. Colussi, S. P. Sander, and R. R. Friedl, J. Phys. Chem. 96, 4442
͑1992͒.
4 N. Basco and S. K. Dogra, Proc. R. Soc. London, Ser. A 323, 401 ͑1971͒.
5 S. P. Sander and R. R. Friedl, J. Phys. Chem. 93, 4764 ͑1989͒.
6 C. M. Nelson, T. A. Moore, and M. Okumura, J. Chem. Phys. 100, 8055
͑1994͒.
22 S. L. Nickolaisen, S. P. Sander, and R. R. Friedl, J. Phys. Chem. ͑submit-
ted͒.
J. Chem. Phys., Vol. 104, No. 8, 22 February 1996
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.42.202.150 On: Mon, 24 Nov 2014 05:04:06