Photochemistry of phosgene in the solid phase: State-resolved dissociation dynamics

Article abstract of DOI:10.1063/1.471506

The translational, rotational, and vibrational state distributions of CO (g) resulting from the single photon photodissociation of Cl₂CO in the condensed phase at ～90 K have been determined by time-of-flight (TOF) distribution measurement and resonance-enhanced multiphoton ionization (REMPI) spectroscopy. The TOF distribution of CO (g) is bimodal. Internal state characterization of the slow channel reveals a completely thermalized origin, with a rotational temperature of T_rot=88±5 K, which is equal to the translational temperature as well as the substrate temperature. We believe these slow CO molecules originate from photodissociation below the topmost surface of the molecular film and achieve thermal equilibrium with the substrate before escaping into the gas phase. Internal state characterization of the fast channel shows, on the other hand, an energetic origin: at hv=5.0 eV, the rotational distribution, with an overall flux-weighted mean rotational energy of 〈E_rot〉=0.12±0.01 eV, is non-Boltzmann and can be approximated by a bimodal distribution with rotational temperatures of 210±40 K at low J″(s) and 2200±300 K at high J″(s); the relative vibrational population is N_v=1/N_v=0=0.33±0.05. Both rotational and translational distributions of fast CO show positive correlation with photon energy. These CO molecules must be promptly ejected into the gas phase, carrying nascent energetic information from the photodissociation reaction on the surface of the molecular film. For electronic excitation events that result in photodissociation, 74% of the excess excitation energy is distributed in the translational and internal motions of products (CO and Cl); only 26% of the available energy is converted to motions of surrounding molecules.

Full text of DOI:10.1063/1.471506

Photochemistry of phosgene in the solid phase: Stateresolved dissociation dynamics
Q.S. Xin and X.Y. Zhu
Citation: The Journal of Chemical Physics 104, 7904 (1996); doi: 10.1063/1.471506
View online: http://dx.doi.org/10.1063/1.471506
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/104/20?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Highresolution infrared spectroscopy of isotopic impurity Q 1(0) transitions in solid parahydrogen
J. Chem. Phys. 105, 4461 (1996); 10.1063/1.472298
Photochemistry of phosgene in the solid phase: Dissociation, ejection, and thermal desorption
J. Chem. Phys. 104, 7895 (1996); 10.1063/1.471505
Stateresolved, threedimensional product recoil velocity spectroscopy
J. Chem. Phys. 104, 2259 (1996); 10.1063/1.470922
Evidence for stepwise dissociation dynamics in acetone at 248 and 193 nm
J. Chem. Phys. 102, 4447 (1995); 10.1063/1.469493
Photodissociation dynamics of the methyl radical 3s Rydberg state
J. Chem. Phys. 102, 792 (1995); 10.1063/1.469193
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.209.6.50 On: Sun, 21 Dec 2014 07:40:45

Photochemistry of phosgene in the solid phase: State-resolved dissociation
dynamics
Q.-S. Xin and X.-Y. Zhua)
Department of Chemistry & Biochemistry, Southern Illinois University, Carbondale, Illinois 62901
͑Received 12 January 1996; accepted 22 February 1996͒
The translational, rotational, and vibrational state distributions of CO (g) resulting from the single
photon photodissociation of Cl2CO in the condensed phase at ϳ90 K have been determined by
time-of-flight ͑TOF͒ distribution measurement and resonance-enhanced multiphoton ionization
͑REMPI͒ spectroscopy. The TOF distribution of CO (g) is bimodal. Internal state characterization
of the slow channel reveals a completely thermalized origin, with a rotational temperature of
Trotϭ88Ϯ5 K, which is equal to the translational temperature as well as the substrate temperature.
We believe these slow CO molecules originate from photodissociation below the topmost surface of
the molecular film and achieve thermal equilibrium with the substrate before escaping into the gas
phase. Internal state characterization of the fast channel shows, on the other hand, an energetic
origin: at h␯ϭ5.0 eV, the rotational distribution, with an overall flux-weighted mean rotational
energy of ͗Erot͘ϭ0.12Ϯ0.01 eV, is non-Boltzmann and can be approximated by a bimodal
distribution with rotational temperatures of 210Ϯ40 K at low JЉ(s) and 2200Ϯ300 K at high JЉ(s);
the relative vibrational population is N␯ϭ1/N␯ϭ0ϭ0.33Ϯ0.05. Both rotational and translational
distributions of fast CO show positive correlation with photon energy. These CO molecules must be
promptly ejected into the gas phase, carrying nascent energetic information from the
photodissociation reaction on the surface of the molecular film. For electronic excitation events that
result in photodissociation, 74% of the excess excitation energy is distributed in the translational and
internal motions of products ͑CO and Cl͒; only 26% of the available energy is converted to motions
of surrounding molecules. © 1996 American Institute of Physics. ͓S0021-9606͑96͒02020-X͔
I. INTRODUCTION
ture of the fast channel therefore provides an excellent op-
portunity to address the detailed dynamics, particularly en-
ergy transfer mechanisms in the condensed state. It is the
purpose of the present study to determine the internal state
distributions of CO (g) resulting from the single photon pho-
todissociation of Cl2CO in the condensed film. The depen-
dence of CO internal state distribution on photon energy as
well as the correlation between translational and internal mo-
tions is also addressed. These measurements are achieved
using state-resolved resonance-enhanced multiphoton ioniza-
tion ͑REMPI͒ spectroscopy.
The state resolved approach is widely used in studying
gas phase reaction dynamics and has also been successfully
applied to the photochemistry of condensed molecules on
cryogenic substrates.3–6 Kutzner et al. achieved partial state-
resolved characterization of CH3 and I from the laser irradia-
tion of condensed CH3I and established a direct photodisso-
ciation mechanism similar to that in the gas phase.3 Natzle
et al. measured the translational and internal state distribu-
tion of NO desorbed from a condensed film.4 The state re-
solved information was crucial in identifying two distinc-
tively different molecular desorption mechanisms: thermal
desorption due to transient laser heating and direct photodis-
sociation due to photoabsorption by ͑NO͒2. Dixon-Warren
et al. studied the photodissociation of NO2 physisorbed on
LiF at multilayer coverages and found that the bimodal vi-
brational state distribution of NO was qualitatively similar to
but quantitatively different from the corresponding process
in the gas phase.5 Buntin et al. measured the final state dis-
As part of our effort to understand photochemical dy-
namics and energy transfer mechanisms in the condensed
phase and, particularly, to compare them to photochemical
process on solid surfaces, we have investigated the photo-
chemistry of Cl2CO in the monolayer region on a Ag
surface1 as well as in a condensed molecular film.2 Our pre-
vious paper on condensed phosgene focused on time-of-
flight ͑TOF͒ measurement. It firmly established that, at the
low laser power density employed ͑10 ␮J–1 mJ/cm2 for ϳ10
ns pulses͒, electronically excited Cl2CO resulting from single
photon photoabsorption in the surface region of the solid film
could either dissociate to form CO (g) and Cl (g) or convert
its electronic energy to translational motion in Cl2CO (g).
The translational energy distribution of CO (g) from the
photodissociation channel was bimodal: the flux-weighted
mean translational energy of the fast channel was photon
energy dependent ͑͗Etrans͘ϭ210, 135, and ϳ90 meV at h␯
ϭ6.4, 5.0, and 3.5 eV, respectively͒, while that of the slow
channel was ͗Etrans/2k͘ϭ84Ϯ3 K, which was independent of
photon energy and was equal to the substrate temperature.
We concluded that the fast channel retained nascent ener-
getic information from the photodissociation reaction on the
surface of the molecular film, while the slow channel was
completely thermalized to the substrate temperature and
likely originated below the topmost surface. The nascent na-
a͒
To whom correspondence should be addressed. Phone: ͑618͒ 453-3077;
Fax: 453-6408; Email: zhu@qm.c-chem.siu.edu
7904
J. Chem. Phys. 104 (20), 22 May 1996
0021-9606/96/104(20)/7904/10/$10.00
© 1996 American Institute of Physics
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
130.209.6.50 On: Sun, 21 Dec 2014 07:40:45

Q.-S. Xin and X.-Y. Zhu: Dissociation dynamics of phosgene
7905
tribution of CO from the photodissociation of Mo͑CO͒6 con-
densed on Si and suggested the importance of collisional
relaxation and energy quenching within the adlayer.6 In the
present study, we focus on the photon energy dependence of
CO final state distributions from the single photon photodis-
sociation of Cl2CO in the condensed film. In particular, we
establish the nascent nature of photodissociation products
and discuss energy partitioning within the molecular film.
The gas phase photochemistry of Cl2CO was summa-
rized by Okabe.7 The Cl2CO molecule in the ground elec-
1
tronic state (X A1) has C2v symmetry. Its UV absorption
spectrum is characterized by a broad continuum between 305
and 238 nm, followed by a stronger absorption band at
shorter wavelength extending to the vacuum ultraviolet. UV
photodissociation of Cl2CO is believed to yield the unstable
ClCO radical, which promptly dissociates to CO and Cl:
FIG. 1. Schematics of the experimental setup.
tion rates of 50 Hz or less. The beam size at the output of the
laser was 1.0ϫ2.5 cm. Since only low laser pulse energy was
needed, we relied on the natural divergence over a long dis-
tance to expand the laser beam to approximately double the
original dimensions. Typically, only the center portion of the
expanded beam with a diameter of ϳ6 mm was directed onto
the sample through a MgF2 vacuum viewport. The laser
pulse energy on the surface was kept at 60 ␮J/cm2. All mea-
surements were carried out at a substrate temperature of ϳ90
K with the photolysis laser at an angle of incidence of 60°
with respect to the surface normal.
Cl2COϩh␯→ClCOϩCl,
ClCO→COϩCl.
͑1͒
͑2͒
The energy requirement for breaking the first Cl–CO bond,
reaction ͑1͒, is 3.3 eV and that for the second Cl–CO bond,
reaction ͑2͒, is only 0.26 eV. The quantum yield for the
photodissociation reaction in the UV region is unity. In the
vacuum ultraviolet, photon absorption by Cl2CO yields elec-
tronically excited Cl2.
II. EXPERIMENT
The probe laser light ͑L2͒ for REMPI was from a tunable
dye laser ͑Continuum ND6000͒ pumped by the second har-
monic of a Q-switched Nd:YAG laser ͑Continuum Powerlite
7010͒. A DCM methanol solution was used to generate tun-
able light at ϳ645 nm. The wavelength was not indepen-
dently calibrated. The dye output ͑ϳ100 mJ͒ was tripled in a
nonlinear optical system ͑Inrad Autotracker II͒. This in-
volved the doubling of the dye output in an angle-tuned KDP
crystal. The polarization of the resulting second harmonic
͑2␻͒, which was orthogonal to the residual fundamental ͑␻͒,
was rotated in a polarization rotator. The two colors ͑␻ and
2␻͒ were then mixed in an angle tuned BBO crystal. The
third harmonic ͑3␻͒ at ϳ215 nm was separated from residual
light in a four-prism wavelength separator, which ensured
that there was no beam-walk during scan. The 215 nm UV
light ͑ϳ2 mJ͒ was focused by a quartz lens ͑fϭ25 cm͒ and
directed into the UHV chamber through a fused silica win-
dow. The probe laser was parallel to the sample and the focal
point 10 mm away from the surface. Note that the size of L1
on the sample was ϳ6 mm. Based on simple geometric ar-
guments, this can introduce a maximum error of Ϯ8% in the
actual flight distance, depending on the geometrical origin of
molecules on the surface.
All experiments were performed in a two-level ultra-
high vacuum ͑UHV͒ chamber, with a base pressure of
1ϫ10Ϫ10 torr. The upper level was used mainly for surface
preparation and analysis. Most experiments were done on the
lower level, which housed a quadrupole mass spectrometer
͑QMS͒ and a time-of-flight mass spectrometer ͑TOFMS͒.
The QMS was used for residual gas analysis ͑RGA͒, thermal
desorption spectroscopy ͑TDS͒, and time-of-flight ͑TOF͒
distribution measurements, as detailed in the previous paper.2
The TOFMS, equipped with Chevron multichannel plates
͑MCP͒ and an impedance matched fast timing anode, was
used for mass resolved ion detection in resonance-enhanced
multiphoton ionization ͑REMPI͒ spectroscopy. In all experi-
ments presented here, 50 monolayers ͑ML͒ of phosgene was
condensed at 90 K on a Ag͑110͒ sample ͑␾10 mm͒, which
was mounted at the center of the chamber. Details on sample
preparation is available in the previous paper.
The experiment, shown schematically in Fig. 1, involved
a classic ‘‘pump–probe’’ scheme: a photolysis laser pulse
͑L1͒ initiates the photochemical reaction on the solid film.
After certain time delay, a probe laser pulse ͑L2͒ intercepts
the products in the gas phase, 10 mm above the substrate.
Molecules within the focal volume of the probe laser are
resonantly ionized and the resulting ions are accelerated to-
ward the TOFMS. After mass separation in a 15 cm drift
tube, the ion signal was detected by the MCPs.
The experiment was completed controlled by a LabView
based software on a personal computer equipped with a tim-
ing I/O board and a data acquisition board. Both the excimer
laser and the YAG laser were triggered at 10 Hz, while the
step motor ͑wavelength scan͒ of the dye laser was triggered
at 1 Hz. The time delay between pump and probe was soft-
ware controlled, with a resolution of 0.2 ␮s. The ions were
mass separated in the TOFMS. Both the ion intensity from
The pump laser ͑L1͒ was a LamdaPhysik Lextra-100 ex-
cimer laser, which generated 193 nm ͑17 ns pulse width͒ and
248 nm ͑23 ns pulse width͒ light when filled with Ar–F2 and
Kr–F2 mixtures, respectively. It could be operated at repeti-
J. Chem. Phys., Vol. 104, No. 20, 22 May 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:
130.209.6.50 On: Sun, 21 Dec 2014 07:40:45

7906
Q.-S. Xin and X.-Y. Zhu: Dissociation dynamics of phosgene
FIG. 2. REMPI spectrum of CO from the photodissociation of condensed Cl2CO at 90 K. The photolysis laser wavelength was 193 nm, with a pulse energy
of 60 ␮J/cm2. The pump–probe delay time was 10 ␮s. The positions of the O, P, Q, R, and S branches for the ͑0,0͒ and ͑1,1͒ hot bands are shown above
and below the spectra, respectively. The x axis is presented on a relative scale, i.e., two photon transition energy (⌬E) minus the band origin ͑E0͒.
the MCPs and probe laser energy were recorded by boxcar
averagers while the photolysis laser power was kept constant
by feedback power locking. The ion intensity was divided by
the square of the simultaneously determined probe laser
power (P), based on a second order power law,8 to correct
for power fluctuations.
served accidental predissociation of the JЈϭ31, 41, and 44
states on the E ⌸, ␯ϭ0 level, as well as the JЈϭ7, 9, 10, and
12 states on the ␯ϭ1 level. Predissociation from these states
1
were found to result from perturbation by various rovibronic
3
levels of the k ⌸ state.
Figure 2 shows a typical REMPI spectrum of CO (g)
obtained from the photodissociation of condensed Cl2CO at a
photolysis laser wavelength of 193 nm. A pump–probe delay
time of 10 ␮s was used in the measurement. As discussed in
detail later, this delay time corresponds to the peak position
of fast CO molecules in the TOF spectrum. Most rotational
lines in the O, P, R, and S branches are well resolved in this
spectrum. The positions of the predicted rotational lines for
the ͑0,0͒ original band, obtained from published rotational
constants,11 are shown above the spectrum. Note that the x
axis is presented on a relative scale, i.e., two photon transi-
tion energy (⌬E) minus the band origin ͑E0͒. The absolute
wavelength from the dye laser output was not calibrated. The
experimental spectra were shifted slightly to match the predi-
cated line positions. All rotational line positions in the ex-
perimental spectra were in excellent agreement with the pre-
dicted positions through out the whole spectrum region.
A detailed analysis of this REMPI spectrum shows that,
in addition to all the peaks from the ͑0,0͒ band, there are a
large number of new peaks from another overlapping band.
These new peaks are in excellent agreement with a predicted
III. RESULTS AND DISCUSSIONS
A. REMPI spectroscopy of CO: Original and hot
bands
Rotationally resolved detection of CO was achieved via
1
1
the E ⌸–X ⌺ϩ transition using a two photon resonance,
three photon ionization scheme, i.e., ͑2ϩ1͒ REMPI.8–10
Hines et al. carried out quantitative analysis of a room tem-
1
1
perature E ⌸–X ⌺ϩ REMPI spectrum.8 They showed that
this scheme provided full rotational resolution and could be
used for the quantitative determination of rotational state dis-
2
tributions. However, the COϩ ion in the A ⌸ state resulting
from this REMPI scheme can absorb an additional photon
from the probe laser and yield predominantly Cϩ and O. This
dissociation channel is significant at high JЉ values in the R
and S branches, due to the relatively high photon energies
required for these resonances. As a result, following the COϩ
ion alone will under-count high JЉ states to some extent.
Recently, Cacciani et al. used this scheme to record a
spectrum of CO at 1000 K and discussed in detail the acci-
1
1
͑1,1͒ vibrational hot band for the E ⌸–X ⌺ϩ two photon
transition. The positions of the predicted rotational lines for
the ͑1,1͒ hot band based on published rotational and vibra-
tional constants10 are shown below the spectrum. Figure 3
shows in detail a portion of the R and S branches of the
spectrum in Fig. 2, along with assignments for the S
branches of the ͑0,0͒ and ͑1,1͒ bands, respectively. For clar-
1
dental dissociation phenomena in the E ⌸ state.10 The E
state lies at 0.29 eV above the CϩO adiabatic dissociation
limit. Accidental predissociation from the E state occurs for
some rotational states on the ␯ϭ0 and 1 vibrational levels
and can significantly reduce the ion intensity of correspond-
ing transitions. They measured the ͑2ϩ1͒ REMPI and the 1
VUVϩ1 UV ionization spectra of CO at 1000 K and ob-
J. Chem. Phys., Vol. 104, No. 20, 22 May 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:
130.209.6.50 On: Sun, 21 Dec 2014 07:40:45

Q.-S. Xin and X.-Y. Zhu: Dissociation dynamics of phosgene
7907
FIG. 3. A portion of the REMPI spectrum in Fig. 2. The assignments are for the S branches of the ͑0,0͒ and ͑1,1͒ bands.
ity, assignments for the R branches are not shown. The ͑1,1͒
J Ϫ2͒J
͑
Љ
Љ
vibrational hot band for the E ⌸–X ⌺ϩ two photon tran-
sition of CO is observed for the first time here. This hot band
was not observed even for CO at 1000 K,10 pointing to the
energetic nature of CO (g) from the photodissociation of
condensed Cl2CO. Note that the intensity of the JЉϭ29 tran-
sition in the S branch is abnormally low. This is due to
1
1
S J ,J ,M͒ϭ
O͒;
͑
͑
Љ Ј
͚
15 2J Ϫ1͒
͑
Љ
M
2
J ϩ1͒
Љ
͑
ϭ
ϭ
ϭ
ϭ
P͒;
͑
30 J Ϫ1͒
͑
Љ
2J ϩ1͒
͑
Љ
Q͒;
͑
1
accidental dissociation of the JЈϭ31 level on the E ⌸ state.
10 2J Ϫ1͒ 2J ϩ3͒
͑
͑
Љ
Љ
The rotational state distribution of CO (g) was obtained
from peak intensities. The overlapping peaks were deconvo-
luted using a least-square fitting routine. The average peak
width was obtained from fitting no-overlapping peaks with a
mixed Gaussian–Lorentzian function. A fixed peak width
was subsequently used in the analysis of all transitions ͑in-
cluding nonoverlapping peaks͒ and the intensity of each tran-
sition was obtained from the fitting curve. Even for well
isolated peaks, we found the error range of ion intensities in
repeated measurements obtained from this fitting procedure
was significantly smaller than that obtained from simple
peak area integration. The ground state rotational population,
N(JЉ), was extracted from the power-normalized peak inten-
sities, I(JЉ,JЈ)/P2, based on the following equation:
J
Љ
R͒;
͑
30
J ϩ3͒ J ϩ1͒
͑
͑
Љ
Љ
S͒.
͑4͒
͑
15 2J ϩ3͒
͑
Љ
Dividing the power normalized peak intensities by the above
line strength yielded N(JЉ)/(2JЉϩ1), which was used in the
so-called Boltzmann plots to yield rotational temperatures.
The severely overlapping rotational lines, as well as lines
associated with accidental predissociation, were discarded in
data analysis.
To demonstrate the reliability of using this REMPI
scheme to obtain rotational state distributions, we show in
the upper panels of Fig. 4 Boltzmann plots of results ob-
tained from a room temperature ͑295 K͒ CO sample. Here,
only the R and S branches are used, since rotational lines are
better resolved in these than in the O and P branches. This is
particularly important for resolving rotational lines from the
͑1,1͒ vibrational hot band of CO from the photodissociation
of phosgene. Results obtained from the O and P branches
were similar. In agreement with the work of Hines et al.,8
following the COϩ signal alone, panel ͑A͒, underestimates
I J ,J ͒
͑
Љ Ј
ϰN J ͒S J ,J ͒,
͑3͒
͑
͑
Љ Ј
Љ
P2
where JЉ is the ground rotational state and JЈ the intermedi-
ate rotational state. S(JЉ,JЈ) is the two photon rotational
1
1
transition strength. For E ⌸–X ⌺ϩ, i.e., ͉⍀ϭ1͘←͉⍀ϭ0͘,
the line strengths summed over all M degenerate states are
given by12
J. Chem. Phys., Vol. 104, No. 20, 22 May 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:
130.209.6.50 On: Sun, 21 Dec 2014 07:40:45

7908
Q.-S. Xin and X.-Y. Zhu: Dissociation dynamics of phosgene
FIG. 5. TOF spectra of CO from the photodissociation of condensed Cl2CO
at 193 and 248 nm. The flight distance was 76 mm. The laser pulse energy
was ϳ10–100 ␮J/cm2. Both spectra were normalized to the number of
incident photons and converted to the flux domain. The dashed curves are
deconvolutions based on modified MB distributions. The solid curves are
the sums of dashed curves. Adapted from the previous paper ͑Ref. 2͒. See
text for details.
FIG. 4. Panel A: Boltzmann plot of the rotational state distribution obtained
from the COϩ ion intensity alone in the REMPI spectrum of a room tem-
perature CO sample ͑Tϭ295 K͒; Panel B: Boltzmann plot of the rotational
state distribution obtained from the COϩϩCϩ ion intensities in the REMPI
spectrum of the room temperature CO sample; Panel C: Boltzmann plot of
the rotational state distribution obtained from the REMPI spectrum of CO at
1000 K. The data points are obtained from peak intensities of a published
spectrum of Cassiani et al. ͑Ref. 10͒. The total ion intensity was followed in
the measurement. In each panel, the open and solid circles correspond to R
and S branches, respectively. The solid lines are least-square linear fits
which yield the indicated rotational temperatures.
equilibrium temperature. The rotational and vibrational state
distributions of CO obtained from the spectrum in Fig. 2 will
be presented later.
To obtain an accurate vibrational population, we need
the relative Franck–Condon factors for the two bands. Both
the rotational temperature by 5%. When intensities from both
COϩ and Cϩ are included in the analysis, panel ͑B͒, the
resulting rotational temperature is in excellent agreement
with the gas temperature. We have also recorded the Cϩ
REMPI spectrum of CO from the photodissociation of phos-
gene ͑not shown͒. However, attempts to extract quantitative
information from the Cϩ spectrum have failed, due to the
low counting level as well as the nonresonant nature of the
Cϩ signal. The origin of this low level nonresonant Cϩ signal
was not known. Since it was not observed for room tempera-
ture CO, we suspect that it might come from the nonresonant
multiphoton ionization of C, resulting from the dissociation
of internally excited CO. In our analysis of CO from photo-
dissociation, the internal state distribution is obtained from
the COϩ signal only, not corrected for by the minor Cϩ
signal.
Analysis of a room temperature CO spectrum only es-
tablishes the reliability of this REMPI scheme for JЉр25.
The validity for extracting rotational state distribution from
ion intensity at higher JЉ levels is shown in panel ͑C͒. The
data points are obtained from the published spectrum of Cac-
ciani et al. for CO at thermal equilibrium at 1000 K.10
Within the uncertainty of our analysis, the rotational tem-
perature, Trϭ950Ϯ90 K, is in excellent agreement with the
the vibrational frequencies ͑␯ ͒ and the equilibrium C–O
0
1
bond distances (re) are very similar for the X ⌺ϩ and the
1
E ⌸ electronic states, indicating that the two FC factors
should be very close. Based on harmonic approximation and
molecular constants in the literature ͓␯0ϭ2143.272 95 cmϪ1
,
1
reϭ1.128 322 Å for the X ⌺ϩ state;11 ␯0ϭ2153.272 95
1
cmϪ1 and reϭ1.121͑9͒ Å for the E ⌸ state13͔, we have cal-
culated the relative FC factors: FC͑1,1͒/FC͑0,0͒ϭ0.98. This
value is taken as unity in the quantification of vibrational
populations.
B. Correlation of translational and internal state
distributions
As detailed in the previous paper1 and discussed earlier,
the translational distribution of CO (g) from the single pho-
ton photodissociation of Cl2CO in the molecular film is bi-
modal. Figure 5 duplicates the time-of-flight spectra of CO at
two photolysis laser wavelength from the previous paper.2
Each spectrum can be deconvoluted into two channels, as
shown by the dashed curves. While the flux-weighted mean
translational energy of the slow channel, ͗Etrans/2k͘ϭ84Ϯ3
K, is independent of h␯, the fast channel is clearly photon
energy dependent ͑͗Etrans͘ϭ210 and 135 meV and h␯ϭ6.4
J. Chem. Phys., Vol. 104, No. 20, 22 May 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:
130.209.6.50 On: Sun, 21 Dec 2014 07:40:45

Q.-S. Xin and X.-Y. Zhu: Dissociation dynamics of phosgene
7909
FIG. 6. REMPI spectra of CO from the photodissociation of condensed Cl2CO at two pump–probe delay times ͑tdelayϭ12 and 38 ␮s͒. The photolysis laser
wavelength was 248 nm, with a pulse energy of 60 ␮J/cm2. The focal point of the probe laser beam was 10 mm away from the surface. The two delay times
correspond to the peak TOF positions in Fig. 5. The positions for the ͑0,0͒ and ͑1,1͒ hot bands are shown above and below the spectra, respectively. The x
axis is presented on a relative scale, i.e., two photon transition energy (⌬E) minus the band origin ͑E0͒.
and 5.0 eV, respectively͒. We suggested in the previous pa-
per that the slow channel corresponded to thermalized CO
originated from below the topmost surface and the fast chan-
nel was due to photodissociation in the surface layer of the
molecular film.
Support for the above interpretation can be found in
REMPI measurements. Figure 6 shows the REMPI spectra of
CO from the 248 nm photodissociation of condensed Cl2CO
at two pump–probe delays. These two delays, 12 and 38 ␮s,
correspond to peak positions for the fast and slow channels,
respectively, in the TOF spectrum ͑Fig. 5͒. Note the different
sample-to-detector distances in TOF and REMPI measure-
ments. In agreement with translation, the REMPI spectra in
Fig. 6 clearly demonstrate the different energetic origins for
the fast and slow channels. While for tdelayϭ38 ␮s rotational
lines up to JЉϭ12 were observed, for tdelayϭ12 ␮s, the ob-
served rotational transitions extend to JЉϭ50.
We have also systematically varied the pump–probe de-
lay times. The results are shown in Fig. 7. Here, only por-
tions of the R and S branches were recorded. Within the slow
TOF peak ͑tdelayϭ26Ϫ100 ␮s͒, the REMPI spectra are iden-
tical when normalized, indicating the same rotational state
distribution. Within experimental uncertainty, the much hot-
ter REMPI spectra in the fast TOF peak are also nearly the
same, pointing to similar internal state distributions.
The rotational state distributions obtained from the
above REMPI spectra are shown in Boltzmann plots in Fig.
8. The results at tdelayϭ26Ϫ100 ␮s can all be well described
by straight lines, yielding rotational temperatures of
Trotϭ88Ϯ5 K, which is identical to the translational tempera-
ture ͑͗Etrans/2k͘ϭ85Ϯ5 K͒ and the substrate temperature
͑ϳ90 K͒. Within experimental uncertainty, Trot is also inde-
pendent of tdelay . Thus, rotational energy in the slow TOF
channel is not correlated with translational energy. All these
results unambiguously establish that CO molecules in the
slow TOF channel are in a complete thermal equilibrium at
the substrate temperature. These CO molecules must result
from photodissociation below the topmost surface of the mo-
lecular film. On their way out to the gas phase, they achieve
thermal equilibrium with the condensed phase.
The results at tdelayϭ8–16 ␮s cannot be described by
straight lines. In other words, these CO molecules are not in
a thermal equilibrium with an effective Boltzmann tempera-
ture. The solid curves at these delays are bimodal fits which
serve as guides to the eye. Details on these rotational state
distributions are discussed in the next section. Within experi-
mental uncertainty, the rotational state distribution of CO
molecules within the fast TOF channel is also independent of
tdelay . Any correlation between translation and rotation, if
present, is below our detection limit, at least for JЉр12. Note
that the rotational distribution at tdelayϭ22 ␮s is a mixed
case, between those at longer and shorter delays. At this
delay time, there is a significant overlap between the fast and
the slow TOF channels.
In the following, we focus on the fast CO channel. These
CO molecules are not thermalized. They retain nascent ener-
getic information from the photodissociation reaction and
should provide insight into the dynamics of this process. The
photon energy dependent translational energy distributions
of photodissociation products have already been discussed in
detail in the previous paper.2 We now address the internal
state distributions.
J. Chem. Phys., Vol. 104, No. 20, 22 May 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:
130.209.6.50 On: Sun, 21 Dec 2014 07:40:45

7910
Q.-S. Xin and X.-Y. Zhu: Dissociation dynamics of phosgene
FIG. 7. R and S branches of the CO REMPI spectra from the photodisso-
ciation of condensed Cl2CO at 248 nm as a function of the indicated pump–
probe delay time ͑tdelayϭ8–100 ␮s͒. The spectra at short delays
͑tdelayϭ8–16 ␮s͒ are scaled by a factor of 2. The focal point of the probe
laser beam was 10 mm away from the sample surface. The x axis is the
probe laser wavelength.
FIG. 8. Boltzmann plots of the rotational state distributions of CO obtained
from the REMPI spectra in Fig. 7. The open and solid circles correspond to
R and S branches, respectively, of the ͑0,0͒ band. The data sets at different
delays are offset for clarity. The solid lines for tdelayϭ26–100 ␮s are least-
squares linear fits which yield the indicated rotational temperatures, while
those for tdelayϭ8–22 ␮s are bimodal fits which serve as guides to the eye
͑see text for detail͒.
C. Internal state distributions of CO and
photodissociation dynamics
actually two sources of CO at two equilibrated rotational
temperatures. The fit is only an approximation to an inher-
ently non-Boltzmann situation. As a matter of fact, despite
the overall quality of the fit, the agreement between the data
The internal state distributions of fast CO molecules
from the photodissociation of condensed Cl2CO at 248 nm
are extracted from the REMPI spectrum in Fig. 6. The results
are shown in Boltzmann plots in Fig. 9. The solid and open
data points are for the ␯ϭ0 and 1 vibrational levels, obtained
from the ͑0,0͒ and ͑1,1͒ bands, respectively. Here, only the R
and S branches are used, since the O and P branches are less
resolved due to the overlap between the ͑0,0͒ and ͑1,1͒
bands. For both vibrational levels, the rotational state distri-
butions are clearly non-Boltzmann-like and cannot be de-
scribed by a single rotational temperature. Each is character-
ized by a fast decay at low JЉ(s) and a slower decay at
higher JЉ(s). To facilitate quantitative comparison, we have
fit each distribution with the sum of two Boltzmann distribu-
tions. The fit ͑solid curve͒ to the ␯ϭ0 vibrational level yields
bimodal rotational temperatures of Trot1ϭ210Ϯ40 K for low
JЉ(s) and Trot2ϭ2200Ϯ300 K for high JЉ(s). The overall
flux-weighted mean rotational energy is ͗Erot͘ϭ0.12Ϯ0.1 eV
or ͗Erot/k͘ϭ1400 K. Due to the limited number of data
points, we did not attempt to fit the data for ␯ϭ1. Within
experimental uncertainty, the data can also be well described
by the same fit as that for the ␯ϭ0 level, indicating similar
rotational state distributions. The relative vibrational popula-
tion is N␯ϭ1/N␯ϭ0ϭ0.33Ϯ0.05. For CO at thermal equilib-
rium, this corresponds to a vibrational temperature of 2800
Ϯ200 K ͑or ͗Evib͘ϭ0.24Ϯ0.02 eV͒.
FIG. 9. Boltzmann plots of the rotational state distributions for fast CO from
the photodissociation of condensed Cl2CO at 248 nm. The data points are
obtained from the REMPI spectrum at tdelayϭ12 ␮s in Fig. 6. The solid
circles and triangles correspond to R and S branches, respectively, of the
͑0,0͒ band and the open circles and triangles are for those of the ͑1,1͒ hot
band. The solid curve on the ␯ϭ0 data is a bimodal fit which yields the
indicated rotational temperatures. The same fitting curve is reproduced and
offset to match the data for ␯ϭ1.
The above fitting procedure does not imply that there are
J. Chem. Phys., Vol. 104, No. 20, 22 May 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:
130.209.6.50 On: Sun, 21 Dec 2014 07:40:45

Q.-S. Xin and X.-Y. Zhu: Dissociation dynamics of phosgene
7911
from the REMPI spectrum in Fig. 2; the solid and open data
points are for the ␯ϭ0 and 1 vibrational levels, respectively.
For both vibrational levels, the rotational state distributions
are again non-Boltzmann-like and can be fit by the sum of
two Boltzmann distributions. The fit ͑solid curve͒ to the ␯ϭ0
vibrational level yields bimodal rotational temperatures of
T
rot1ϭ360Ϯ60 K for low JЉ(s) and Trot2ϭ2200Ϯ300 K for
high JЉ(s). The overall flux-weighted mean rotational en-
ergy is ͗Erot͘ϭ0.13Ϯ0.01 eV or ͗Erot/k͘ϭ1500Ϯ100 K. The
fit for ␯ϭ0 is also shown on the data for ␯ϭ1 to indicate the
similar rotational state distribution. The relative vibrational
population is N␯ϭ1/N␯ϭ0ϭ0.33Ϯ0.05, same as that at 248
nm. A comparison of Figs. 9 and 10 reveals that the differ-
ence between the rotational state distributions at the two pho-
ton energies is negligible at high JЉ(s) but significant at low
JЉ(s). These results contradict the collisional cooling mecha-
nism, which predicts that the distribution at low J(s) should
be more insensitive to h␯ than that at high JЉ(s). Therefore,
our logical conclusion is that the results in Figs. 9 and 10
represent nascent rotational state distributions from the pho-
todissociation reaction, without significant contribution from
those CO molecules which have suffered collisional cooling
before leaving the surface of the molecular film. Note that
collision between CO molecules in the gas phase after leav-
ing the surface is not important here. The low laser power
density guarantees that the transient density of gas phase
products is several orders of magnitude lower than that nec-
essary for significant collision in the gas phase.
The above results establish the nascent nature of CO
final state distributions. This allows us to examine the de-
tailed dynamics, particularly energy partitioning, in the pho-
todissociation reaction. Here, we focus on electronic excita-
tion events which lead to dissociation. Those which do not
lead to photodissociation can result in molecular ejection
and/or thermal motion, as discussed in detail in the previous
paper. Table I summarizes the flux-weighted mean energies
of fast CO (g) from the photodissociation of condensed
Cl2CO at 90 K. At both photon energies, the nascent CO
clearly exhibits disparate translational, rotational, and vibra-
tional energies. This result, along with the non-Boltzmann
rotational state distribution of CO, rules out the possibility of
any slow photodissociation mechanism, such as that implied
by the ‘‘statistical model’’ in which the product final state
distribution is mainly governed by density of states.7 Any
slow photodissociation process, on the time scale of several
vibrational periods, should inevitably lead to substantial loss
FIG. 10. Boltzmann plots of the rotational state distributions for fast CO
from the photodissociation of condensed Cl2CO at 193 nm. The data points
are obtained from the REMPI spectrum in Fig. 2. The solid circles and
triangles correspond to R and S branches, respectively, of the ͑0,0͒ band and
the open circles and triangles are for those of the ͑1,1͒ hot band. The solid
curve on the ␯ϭ0 data is a bimodal fit which yields the indicated rotational
temperatures. The same fitting curve is reproduced and offset to match the
data for ␯ϭ1.
points and the solid curve at low JЉ(s) is not satisfactory. As
shown in more detail by the data at tdelayϭ8–16 ␮s in Fig. 8,
the distribution at low JЉ(s) is also not linear and cannot be
satisfactorily described by a single rotational temperature.
A natural interpretation for the data in Fig. 9 is that the
nascent CO molecules from the photodissociation reaction
on the molecular film is rotationally hot and can be approxi-
mated by an effective rotational temperature of ϳ2200 K.
Due to orientational or other morphological effects, some
CO molecules are not ejected promptly into the gas phase
but rather go through one or more collisions with neighbor-
ing cold molecules. The loss of energy ͑cooling͒ in these
collisional processes can lead to the enhanced flux at low
JЉ(s) in Fig. 9.
Despite the attractiveness of the above proposal, it fails
to account for the dependence of rotational state distribution
on photon energy. Figure 10 shows the internal state distri-
butions of fast CO molecules from the photodissociation of
condensed Cl2CO at 193 nm. The data points are extracted
TABLE I. Flux-weighted mean energies of nascent CO from the phodissociation of condensed Cl2CO.
Translation
͑meV͒
Rotation
Vibration
␭
͑nm͒
h␯
͑eV͒
a
b
͑K͒
͑meV͒
͑K͒
͑meV͒
͑K͒
193
248
6.4
5.0
210Ϯ20
135Ϯ5
1200Ϯ100
130Ϯ10
360Ϯ60,
2200Ϯ300
210Ϯ40,
240Ϯ20
2800Ϯ200
2800Ϯ200
790Ϯ30
120Ϯ10
240Ϯ20
2200Ϯ300
aTemperatures from bimodal fits; see text for detail.
bEquivalent temperature at thermal equilibrium to give N␯ϭ1/N␯ϭ0ϭ0.33Ϯ0.05.
J. Chem. Phys., Vol. 104, No. 20, 22 May 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:
130.209.6.50 On: Sun, 21 Dec 2014 07:40:45

7912
Q.-S. Xin and X.-Y. Zhu: Dissociation dynamics of phosgene
of excess excitation energy to the cold environment, which is
not observed here. This can be demonstrated by simple en-
ergetic considerations. For a photodissociation process de-
scribed by the following overall reaction,
tainty, we found no correlation between translation and rota-
tion ͑Fig. 8͒ or between rotation and vibration ͑Figs. 9 and
10͒.
The dependence of product state distribution on photon
energy provides particular insight into dissociation dynam-
ics. Among the three coordinates of motion, ͗Etrans͘ depends
Cl2COϩh␯→COϩ2Cl,
͑5͒
most strongly on h␯, while ͗Evib͘ or N␯ϭ1/N
is, within
the maximum excess energy available to the products is sim-
ply the difference between the photon energy and the energy
required to break two Cl–CO bonds ͑3.56 eV͒. At h␯ϭ5.0
eV, the available energy is Eaϭ1.44 eV; the total energy
␯ϭ0
experimental uncertainty, independent of h␯. The rotational
state distribution is an intermediate situation, in which the
distribution at low JЉ(s) is h␯ dependent while that at high
JЉ(s) is insensitive to h␯. The insensitivity of CO vibra-
tional distribution suggests the dominance of a mechanism
implied by the ‘‘equilibrium geometry’’ model.7 To elabo-
rate, the vibrational distribution of product CO is determined
by the C–O bond length at the transition state for Cl–CO
bond cleavage. The C–O bond in ClCO is likely more
stretched than that in CO. As a result, vibrational excitation
in CO is expected from this reaction. The positive correlation
of ͗Etrans͘ and h␯ can be interpreted by a simple ‘‘impulsive
model.’’ 7 In this simple model, the initial photoexcitation
provides a sudden impulse to one of the two Cl–CO bonds.
This ‘‘impulse,’’ which scales with h␯, is converted to trans-
lational energies in Cl and ClCO and the internal energy in
Cl–CO, based on momentum considerations. The internal
energy in Cl–CO is in turn converted to motions of CO and
Cl. The rotational distribution can perhaps be interpreted by
a combination of ‘‘equilibrium geometry’’ and ‘‘sudden im-
pulse’’ mechanisms. We must point out that the above dis-
cussions are only speculations based on models developed in
the gas phase for extreme situations. A microscopic picture
of how surrounding molecules affect the photochemical dy-
namics in the condensed phase remains to be developed. In
this regard, detailed knowledge on the state resolved dynam-
ics of corresponding gas phase photochemistry would be
rather beneficial.
distributed in the products is ͗Etotal͘ϭ͗E
trans͘COϩ͗Erot͘CO
ϩ͗Evib COϩ2͗Etrans͘Clϭ1.06 eV. Therefore, only 26% of the
͘
available energy is lost to the environment. Such a small
percentage of energy loss is remarkable, considering the fact
that the dissociating Cl2CO molecule is embedded in a cold
matrix of other Cl2CO molecules. The nascent products must
be promptly ejected into the gas phase; the small energy loss
perhaps only reflects the immediate movement of neighbor-
ing molecules during the dynamic process. We are not able
to reliably address the energetics at h␯ϭ6.4 eV, due to the
lack of accurate knowledge on ͗Etrans͘ of Cl at this photon
energy. We estimate that the percentage of energy loss to the
environment is higher, ϳ50%. Whether radiative energy loss
or electronically excited Cl is also involved remains an open
question.
A possible pathway for prompt energy transfer from the
dissociating molecule to the surrounding can be compre-
hended if we borrow the mechanistic picture on ‘‘cage ef-
fect,’’ developed for photodissociation in the solution
phase.14 If a Cl2CO molecule absorbs a photon, it is pro-
moted to the dissociative excited potential, which in the gas
phase should lead to dissociation with 100% quantum yield.7
On the surface of the molecular film, the separating frag-
ments on the excited potential should encounter surrounding
molecules within the first molecular layer, as well as second
layer molecules immediately below, thus forcing a transient
‘‘expansion’’ of the local environment. This can serve as an
efficient pathway for prompt energy transfer in dissociation
dynamics.
Another possible mechanism for energy loss to the envi-
ronment is via inter-molecular quenching of the electroni-
cally excited state. Depending on the location on the reaction
coordinate where quenching occurs ͑in other words, the life-
time on the excited state͒, the electronically quenched mol-
ecule may still proceed to dissociation, albeit with less en-
ergy available to products. In this scenario, the shorter time
the molecule spends on the excited state, the less energy
͑provided it meets the minimum energetic requirement͒ is
available to the products. A similar mechanism, the so-called
Menzel–Gomer–Redhead ͑MGR͒ model, has been well es-
tablished for photodissociation on solid surfaces.15 However,
we believe such an electronic quenching process is not im-
portant for the final state distributions here. Quenching of
electronic energy should be manifested in the decrease in
energy disposal to the products’ freedom of motions. There-
fore, a correlation between various degrees of freedom,
which reflects the distribution of lifetime on the excited po-
tential, is usually expected. Within our experimental uncer-
IV. CONCLUSIONS
The final state distributions of CO (g) resulting from the
single photon photodissociation of Cl2CO in a solid film are
characterized by two distinctively different channels: ͑i͒ a
completely thermalized channel which can be attributed to
photodissociation products originating below the topmost
surface; and ͑ii͒ an energetic channel with disparate and pho-
ton energy dependent final state distributions. CO molecules
from this channel must be promptly ejected into the gas
phase, carrying nascent energetic information from the pho-
todissociation reaction on the surface of the molecular film.
For electronic excitation events that result in photodissocia-
tion at h␯ϭ5.0 eV, 74% of the excess excitation energy is
distributed in the translational and internal motions of prod-
ucts ͑CO and Cl͒.
ACKNOWLEDGMENTS
This work was supported by the National Science Foun-
dation, Grant No. CHE-9415337. Partial support from the
Camille and Henry Dreyfus New Faculty Award is also ac-
knowledged.
J. Chem. Phys., Vol. 104, No. 20, 22 May 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:
130.209.6.50 On: Sun, 21 Dec 2014 07:40:45

Q.-S. Xin and X.-Y. Zhu: Dissociation dynamics of phosgene
7913
1 Q.-S. Xin and X.-Y. Zhu, J. Chem. Phys. 104, 4274 ͑1996͒.
2 Q.-S. Xin and X.-Y. Zhu, J. Chem. Phys. 104, 7895 ͑1996͒.
3 J. Kutzner, G. Lindeke, K. H. Welge, and D. Feldmann, J. Chem. Phys.
90, 548 ͑1989͒.
8 M. A. Hines, H. A. Michelsen, and R. N. Zare, J. Chem. Phys. 93, 8557
͑1990͒.
9 A. Fujii, T. Ebata, and M. Ito, Chem. Phys. Lett. 161, 93 ͑1989͒.
10 P. Cacciani, W. Hogervorst, and W. Ubachs, J. Chem. Phys. 102, 8308
͑1995͒.
4 W. C. Natzle, D. Padowitz, and S. J. Sibener, J. Chem. Phys. 88, 7975
͑1988͒.
11 P. H. Krupenie, The Band Structure of Carbon Monoxide ͑NBS, Wash-
ington, 1966͒.
5 St. J. Dixon-Warren, R. C. Jackson, J. C. Polanyi, H. Reiley, J. G. Shapter,
and H. Weiss, J. Phys. Chem. 96, 10983 ͑1992͒.
6 S. A. Buntin, R. R. Cavanagh, and L. J. Richter, J. Chem. Phys. 98, 7651
͑1993͒.
12 R. G. Bray and R. M. Hochstrasser, Mol. Phys. 31, 1199 ͑1976͒.
13 S. G. Tilford and J. D. Simmons, J. Phys. Chem. Ref. Data 1, 147 ͑1972͒.
14 A. L. Harris, J. K. Brown, and C. B. Harris, Annu. Rev. Phys. Chem. 39,
341 ͑1988͒.
7 H. Okabe, Photochemistry of Small Molecules ͑Wiley, New York, 1978͒.
15 X.-Y. Zhu, Annu. Rev. Phys. Chem. 45, 113 ͑1994͒.
J. Chem. Phys., Vol. 104, No. 20, 22 May 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:
130.209.6.50 On: Sun, 21 Dec 2014 07:40:45