Femtosecond dynamics of photoinduced molecular detachment from halogenated alkanes. I. Transition state dynamics and product channel coherence

Article abstract of DOI:10.1063/1.477045

The direct observation of the photoinduced molecular detachment of halogens X₂ from halogenated alkanes RCHX₂ is presented. Three-photon excitation at 312 nm produces molecular halogens and a carbene; the halogen products are formed predominantly in the D^′ state. Femtosecond pump-probe spectroscopy of the reaction reveals a fast (τ<50fs) dissociation with no evidence of intramolecular vibrational redistribution. This is consistent with a prompt dissociation without intermediates. The experimental results demonstrate vibrational coherence in the halogen product, which requires that the reaction proceed by a concerted mechanism.

Full text of DOI:10.1063/1.477045

Femtosecond dynamics of photoinduced molecular detachment from halogenated
alkanes. I. Transition state dynamics and product channel coherence
Una Marvet, Qingguo Zhang, Emily J. Brown, and Marcos Dantus
Citation: The Journal of Chemical Physics 109, 4415 (1998); doi: 10.1063/1.477045
View online: http://dx.doi.org/10.1063/1.477045
View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/109/11?ver=pdfcov
Published by the AIP Publishing
Articles you may be interested in
Femtosecond absorption study of photodissociation of diphenylcyclopropenone in solution:Reaction dynamics
and coherent nuclear motion
J. Chem. Phys. 120, 4768 (2004); 10.1063/1.1645778
Photodissociation dynamics of acetylacetone: The OH product state distribution
J. Chem. Phys. 110, 11850 (1999); 10.1063/1.479126
Chlorine atom formation dynamics in the dissociation of CH 3 CF 2 Cl(HCFC-142b) after UV laser
photoexcitation
J. Chem. Phys. 110, 11823 (1999); 10.1063/1.479124
Femtosecond dynamics of photoinduced molecular detachment from halogenated alkanes. II. Asynchronous
concerted elimination of I 2 from CH 2 I 2
J. Chem. Phys. 109, 4428 (1998); 10.1063/1.477046
H atom formation dynamics in the dissociation of CH 3 –CF 2 Cl (HCFC-142b) after UV and VUV laser
photoexcitation
J. Chem. Phys. 107, 779 (1997); 10.1063/1.474376
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:
128.143.23.241 On: Mon, 22 Dec 2014 20:33:30

JOURNAL OF CHEMICAL PHYSICS
VOLUME 109, NUMBER 11
15 SEPTEMBER 1998
Femtosecond dynamics of photoinduced molecular detachment
from halogenated alkanes. I. Transition state dynamics and product
channel coherence
Una Marvet, Qingguo Zhang,a) Emily J. Brown, and Marcos Dantusb)
Department of Chemistry and Center for Fundamental Materials Research, Michigan State University,
East Lansing, Michigan 48824-1322
͑Received 30 January 1998; accepted 12 June 1998͒
The direct observation of the photoinduced molecular detachment of halogens X2 from halogenated
alkanes RCHX2 is presented. Three-photon excitation at 312 nm produces molecular halogens and
a carbene; the halogen products are formed predominantly in the D state. Femtosecond pump–
Ј
probe spectroscopy of the reaction reveals a fast (␶Ͻ50 fs) dissociation with no evidence of
intramolecular vibrational redistribution. This is consistent with a prompt dissociation without
intermediates. The experimental results demonstrate vibrational coherence in the halogen product,
which requires that the reaction proceed by a concerted mechanism. © 1998 American Institute of
Physics. ͓S0021-9606͑98͒02335-6͔
I. INTRODUCTION
halogen bonds break is different. Again, the interhalogen
bond forms before the halogens are completely dissociated
from the carbene. In this paper, the femtosecond transition
state dynamics and vibrational coherence in the halogen
product will be studied in detail. In the companion article
͑referred to as paper II henceforth͒, the rotational anisotropy
of the halogen product will be used to identify the mecha-
nism of one dissociation pathway leading to the formation of
I2.9
Under high energy excitation, several simple polyatomic
molecules have been known to dissociate to yield molecular
photofragments, e.g., H2O→H2ϩO, NH3→NHϩH2,
H2CO→H2ϩCO ͑Ref. 1͒ and COCl2→Cl2ϩCO ͑Ref. 2͒. In-
stead of breaking a single bond as in the majority of photo-
dissociation processes studied to date, this photoinduced mo-
lecular detachment or molecular photodetachment process
involves breakage and formation of more than one bond.
Several dihaloalkanes, most notably CH2IBr ͑Ref. 3͒ and
CH2I2 ͑Refs. 4–8͒, have also been known to undergo mo-
lecular detachment processes. As this study demonstrates,
many other gem-dihaloalkanes, e.g., CH2Br2, CH2Cl2 and
n-C4H8I2, also photodissociate to yield dihalogen molecules
as photofragments.
The spectroscopy and photodissociation dynamics of
CH2I2 have received a great deal of attention over the past
two decades.6,7,10–24 Most of these studies are concerned
with the photodissociation dynamics of low-lying excited
electronic states,10–21 which dissociate to produce CH2I and I
2
fragments with I in either its ground P3/2 state or its spin-
2
orbit excited P1/2 state. However, upon high energy excita-
This paper is the first of a series in which the time-
resolved dynamics of photoinduced molecular detachment
processes will be examined. The reaction of interest to this
study is the molecular detachment of X2 from RCHX2,
tion, CH2I2 has been known to undergo molecular detach-
ment to yield I2 in highly excited electronic states, i.e.,
CH2I2→CH2ϩI2. Following the first observation of this
pathway by Style and co-workers,4,5 Black8 and Okabe
et al.7 conducted thorough spectroscopic investigations of
the photodissociation process. Black showed that the I2 342
nm fluorescence intensity reaches its maximum value within
10 ns of excitation, which is considerably less than 50 ␮s,
the mean collision time under their experimental conditions.8
This suggests that I2 formation is a primary event. Although
the quantum yield measurements for I2 342 nm fluorescence
RCHX ϩh →RCHϩX
v
2
2
͑where XϭCl, Br or I͒. Possible mechanisms for this reac-
tion are shown schematically in Fig. 1. In ͑a͒ the two
carbon–halogen bonds break sequentially and X2 forms later
by a collisional process. This is a stepwise mechanism; the
molecular halogen is a secondary product. The pathway
shown in Fig. 1͑b͒ is concerted, i.e., it proceeds in a single
kinetic step. Breaking of the two carbon–halogen bonds is
initiated at the same time and proceeds at the same rate. An
interhalogen bond forms before the carbon–halogen bonds
are completely broken. In the asynchronous concerted
mechanism ͑c͒ the reaction also proceeds in a single kinetic
step. In this case, however, the rate at which the two carbon–
8
determined by Black (␾fϭ0.002, excitation at 121.6 nm͒
differ from those of Okabe et al. (␾fϭ0.006, excitation at
123.6 nm͒,7 they both indicate that the molecular detachment
process represents only a minor relaxation pathway for the
photoexcitation. From analysis of the dispersed fluorescence
spectra, Okabe et al. also found moderately high vibrational
excitation ( up to 35͒ in the nascent I fragment.
v
Ј
2
The short time dynamics of CH2I2 has been studied re-
cently by many groups using both femtosecond time-
resolved techniques19,20 and resonance Raman spectroscopic
a͒
Current address: George R. Harrison Spectroscopy Laboratory, Massachu-
setts Institute of Technology, Cambridge, MA 02139.
b͒
Author to whom correspondence should be addressed.
0021-9606/98/109(11)/4415/13/$15.00
4415
© 1998 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:
128.143.23.241 On: Mon, 22 Dec 2014 20:33:30

4416
J. Chem. Phys., Vol. 109, No. 11, 15 September 1998
Marvet et al.
CH2Br2 and CH2Cl2 will be presented in Sec. III. In Sec. IV,
detailed analysis of the fluorescence spectra, the transition
state dynamics and the photofragment dynamics will be pre-
sented. The experimental data form the basis for a discussion
of the molecular detachment mechanism of this reaction.
II. EXPERIMENT
The laser system used for these experiments is a home
built colliding pulse mode-locked dye laser ͑CPM͒, pumped
by an Arϩ laser. The output from the CPM is centered at 624
nm and is amplified using a four-stage dye amplifier pumped
by a 30 Hz Nd-YAG laser. A double-pass prism pair recom-
presses the pulses after amplification to produce 50 fs pulses
with an average energy of 0.5 mJ per pulse. The pulses were
characterized using a frequency-resolved optical gating
͑FROG͒ method.
Pump and probe pulses were prepared using a Mach-
Zehnder interferometer. A 0.1 mm KDP second harmonic
generation ͑SHG͒ crystal in the fixed arm was used to pro-
duce the 312 nm pump pulses. The two beams were col-
linearly recombined using a dichroic beam splitter and fo-
cused into the sample cell. Laser induced fluorescence ͑LIF͒
was detected perpendicular to the laser path using a 0.27 m
monochromator and a photomultiplier tube ͑PMT͒. The sig-
nal was collected by a boxcar integrator and stored in a com-
puter for analysis. Spectra were wavelength calibrated using
mercury lamp emission.
All experiments were performed on neat vapor
͑ϳ0.1–10 Torr͒ of the relevant alkyl halide in a static quartz
cell ͑ice baths were used when necessary͒. The purity of the
samples was checked by nuclear magnetic resonance ͑NMR͒
and gas chromatography ͑GC͒. Dehydrated sodium thiosul-
fate was used as a scavenger to prevent the accumulation of
molecular halogens. The 312 nm ͑pump͒ pulse initiates the
reaction by multiphoton excitation of the sample vapor; dis-
sociation products are detected by fluorescence. To obtain
time-resolved data, the probe pulse at 624 nm was used to
deplete the population of halogen molecules in the fluores-
cent state. Sweeping the time delay between the pump and
probe pulses thus yields a transient that reflects the temporal
evolution of the reaction. At each pump–probe time delay,
the signal is collected for 10 laser shots; pulses with energy
more than one standard deviation from the mean are dis-
carded. Typical transients contain data from 200 different
time delays and are averages of 100 scans.
FIG. 1. Schematic of possible mechanisms for the photoinduced molecular
detachment of X2 from gem–dihaloalkanes. The time scales given for these
mechanisms are based on our time-resolved measurements. Three processes
are depicted: ͑a͒ Stepwise ͑nonconcerted͒ mechanism; one halogen–carbon
bond breaks some time before the other. Halogen products are formed later,
by collisional association. M represents a third body in the collision process,
which is required for formation of a diatomic. No coherence is expected
from this pathway. ͑b͒ In the synchronous concerted mechanism, breaking
of the two carbon–halogen bonds is initiated at the same time and proceeds
at the same rate. An interhalogen bond forms before the carbon–halogen
bonds are completely broken. This pathway has a symmetric transition state
and preserves the C2v symmetry of the parent. This mechanism would be
expected to produce molecular halogen products with a small rotational
angular momentum and a significant degree of vibrational excitation. ͑c͒ In
the asynchronous concerted mechanism the reaction proceeds in a single
kinetic step, as described above. In this case, however, the rate at which the
two carbon–halogen bonds break is different. Again, the interhalogen bond
forms before the halogens are completely dissociated from the carbene. This
mechanism produces halogen molecules with a large rotational angular mo-
mentum.
measurements.15–18 However, all these studies are concerned
with the photodissociation dynamics for breaking of a single
C-I bond. It was found that the dissociation process takes
place in less than 120 fs19,20 and that the I-C-I symmetric
stretching mode is initially activated, which leads to breaking
of one of the two C-I bonds by coupling to the I-C-I anti-
symmetric stretching mode.15–18
A series of efforts has been made in our group to inves-
tigate the molecular photodetachment dynamics of CH2I2
and its derivatives.9,25–28 The existence of coherent vibra-
tional motion in the I2 photofragment has been observed.
This is an indication that the molecular photodetachment
process is concerted, i.e., the breaking of the two C-I bonds
and the formation of the I-I bond occur in a single kinetic
step.
III. RESULTS
The dispersed fluorescence spectra resulting from multi-
photon excitation of CH2I2, CH2Br2 and CH2Cl2 at 312 nm
are presented in Fig. 2. The fluorescence intensity has not
been corrected for detection efficiency of the spectrometer.
Most of the spectral features can be assigned to fluorescence
from nascent halogen molecules resulting from photodisso-
ciation of the parent. One of the most striking observations is
that in all cases the principal fluorescent product is a halogen
This study and paper II9 represent a continuation of our
effort to elucidate detailed photodissociation dynamics of
gem-dihaloalkanes using femtosecond pump–probe tech-
niques. The experimental setup will be described in Sec. II of
this paper. Dispersed fluorescence spectra and time-resolved
transients following multiphoton excitation of CH2I2,
molecule in the D state, which is an ion-pair state, i.e., it
Ј
correlates to XϩϩXϪ.29–34 In the case of CH2I2, this is con-
sistent with frequency-resolved studies of photodissociation
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:
128.143.23.241 On: Mon, 22 Dec 2014 20:33:30

J. Chem. Phys., Vol. 109, No. 11, 15 September 1998
Marvet et al.
4417
FIG. 2. Fluorescence spectra resulting from the multiphoton dissociation of
gem–dihaloalkanes. All spectra were obtained by gas-phase excitation at
312 nm in a static cell. ͑a͒ From CH2I2. The dominant fluorescence is
between 300 and 350 nm and can be assigned to the I D →A transition. A
FIG. 3. Dispersed fluorescence spectra from the D state of I . ͑a͒ Fit to the
Ј
2
I D →A spectrum produced from the dissociation of CH I . The top panel
Ј
Ј
Ј
Ј
2
2 2
2
shows the fit using only D →A fluorescence; in the lower panel, fluores-
Ј
Ј
small amount of laser scatter is detected at 312 nm ͑shaded͒. The region
marked ͑i͒ indicates fluorescence produced by other ion-pair states of I2; this
region will be investigated in detail in paper ͑II͒. ͑b͒ From CH2Br2, showing
the D →A fluorescence of Br2. ͑c͒ From CH2Cl2; again, a characteristic
cence due to the f→B transition was also taken into account ͑see text͒. ͑b͒
Fit to a collisionally relaxed ͑Boltzmann͒ distribution of vibrational levels of
the D state. The data were obtained from a cell containing I vapor and Ar
Ј
Ј
Ј
2
gas at 100 Torr.
D →A fluorescence is observed, this time due to Cl2. In this case the
Ј
Ј
signal is weak. The increased signal from 280 toward 310 nm is due to laser
scatter.
at high energies.4–7 The I D →A fluorescence is well char-
presence of buffer gases.29–31 Under these circumstances the
Ј
Ј
expect to see fluorescence in the 300–350 nm region of the
spectrum, assignable to the f→B transition. Figure 3͑a͒
shows the fit to the spectrum of the I2 fluorescence produced
in this region by photodissociation of CH2I2. The fit shown
in the top panel in Fig. 3͑a͒ was obtained by assuming that
all the fluorescence in this region was produced by the I2
(D →A ) transition; the second panel shows the same data
2
acterized and is commonly observed when I2 is excited in the
D
state is populated by collisional relaxation from other
Ј
ion-pair states; this process is very efficient. However, the
pressure in the reaction cell is ϳ1 Torr, which precludes the
possibility of forming I (D ) by collisional relaxation, since
Ј
Ј
Ј
2
the time between collisions at this pressure, Ϸ130 ns, is
much longer than the fluorescence lifetime of other ion-pair
states ͑ϳ10 ns͒.35 In addition, a collisional formation process
would produce a rise time on the order of 100 ns. The ob-
served signal had a very rapid rise time and was collected
fit with Ϸ30% ͑optimized for fit͒ of f→B fluorescence. Al-
though there are still deviations from the fit, the second panel
shows a better correspondence to the data than in the first
panel, particularly in the region between 322 and 337 nm.
Remaining differences may be due to errors in the assumed
vibrational distribution in the f state; we will investigate this
further. The fits were both obtained by using a least-squares
routine and by assuming a Gaussian distribution of vibra-
with a boxcar integrator having a 50 ns gate. The D state of
Ј
halogens is optically inaccessible from the ground state; ob-
servation of this transition therefore indicates that the I2 fluo-
rescence is originating from a nascent dissociation product.
It can be seen from the CH2I2 spectrum that there are
also fluorescence bands between 250 and 290 nm; the
f(3⌸0ϩg)→A(3⌸1u) ͑Ref. 36͒, F(1⌺ϩ0u)→X(1⌺0ϩg) ͑Ref. 36͒,
tional levels in the D state; the observed fits correspond to a
Ј
central vibrational level maxϭ9.
v
The D →A fluorescence that would be observed from
Ј
Ј
collisionally relaxed I (D ) molecules was also simulated;
Ј
2
ϩ
ϩ
Ϫ
Ϫ
3
f (1⌺0g)→B(3⌸0u) ͑Ref. 37͒, g(0g )→ ⌸(0u ) ͑Ref. 36͒
Fig. 3͑b͒ shows a comparison of the simulation with the
dispersed fluorescence obtained from a sample of I2 in 100
Torr of Ar at ambient temperature. This is expected to pro-
Ј
and G(1g)→A(3⌸1u) ͑Ref. 36͒ transitions of I2 fluoresce in
this region. The integrated intensity of the fluorescence in the
250–290 nm region relative to the 290–350 nm region is
approximately 10%. A detailed analysis of the signal
observed in this region is presented in paper II.9
duce a Boltzmann population of vibrational states in the D
Ј
electronic state of I2. Figure 3͑b͒ shows that the match be-
tween the simulated spectrum and the experimental data is
If the f(3⌸0ϩg) state of I2 were populated, we would also
quite good. It is clear from Figs. 3͑a͒ and 3͑b͒ that the I (D )
Ј
2
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:
128.143.23.241 On: Mon, 22 Dec 2014 20:33:30

4418
J. Chem. Phys., Vol. 109, No. 11, 15 September 1998
Marvet et al.
FIG. 5. Excitation schemes for I2 and CH2I2 showing how observed time-
dependent fluorescence is generated. ͑I.P. refers to the ionization potential of
the molecule.͒ ͑a͒ From I2 vapor ͑neat͒. The 624 nm ͑pump͒ pulse excites I2
to the B state. Subsequently, a 312 nm pulse promotes the excited I2 mol-
ecules to the E and f states; fluorescence from the f→B transition is de-
tected at 340 nm. The vibrational oscillations observed in Fig. 4͑a͒ are
produced by Franck–Condon transitions from the wave packet oscillating
on the B state potential energy surface. ͑b͒ From CH2I2 vapor. Initially, the
molecule undergoes multiphoton absorption of the 312 nm ͑pump͒ pulse.
FIG. 4. Time-resolved data of I2 ͑refer also to Fig. 5͒. ͑a͒ From iodine vapor
͑neat͒, showing no signal at negative times and vibrational coherence char-
acteristic of I2(B) at positive times. A 624 nm ͑pump͒ pulse excites I2 to the
B state; subsequent absorption of a 312 nm pulse causes transition to the E
and f states; signal is f→B fluorescence at 340 nm. ͑b͒ From the dissocia-
tion of CH2I2. Multiphoton absorption of 312 nm photons causes photoin-
duced molecular detachment of I2 in the D state. At positive times, fluo-
Ј
rescence at 340 nm from the D →A transition is depleted by the 624 nm
Ј
Ј
This causes it to dissociate, producing I in the D state. Absorption of a 624
͑probe͒ pulse. Observation of vibrational modulation in the depletion signal
Ј
2
nm ͑probe͒ pulse by the iodine depletes the population of the D state.
indicates that a coherent wave packet is generated in the D state of I , see
Ј
Ј
2
Depletion efficiency tracks the progress of the reaction and is monitored by
Fig. 5͑b͒. Note that for the I2 transient, positive time corresponds to the 624
nm pulse arriving at the cell first; for the CH2I2 sample, positive time is
when the 312 nm pulse arrives first.
detecting the D →A fluorescence at 340 nm. Note that the probe transition
Ј
Ј
may not be as represented; there are a number of optically accessible disso-
ciative I states 2 eV below the D state.
Ј
2
population resulting from the photodissociation of CH2I2 can
not be represented by a Boltzmann distribution.
sively studied and produces vibrational oscillations charac-
teristic of the B state of I2.38
To further confirm that the observed fluorescence signal
does not originate from background I2 in the cell, time-
resolved data of the fluorescence signal at 340 nm were col-
lected for two cells, each containing CH2I2 or I2. Figure 4͑a͒
shows the results from the I2 cell. When I2 absorbs a photon
of energy 2 eV ͑provided by the 624 nm pulse͒, it is excited
to the B state as shown in Fig. 5͑a͒. Absorption of an addi-
tional 4 eV from the 312 nm pulse causes excitation to the
ion-pair states E and f, which are fluorescent. The transient
obtained by detecting the 340 nm fluorescence from the
f→B transition is shown in Fig. 4͑a͒. Because ground-state
I2 does not absorb at 4 eV, no fluorescence signal is observed
from the iodine cell when the 312 nm pulse arrives first
͑negative time͒. As the pulses become overlapped in time
͑time zero͒, the ion-pair states are populated and the fluores-
cent signal begins to appear. At positive times ͑624 nm pulse
arriving first͒, modulation in the fluorescence intensity can
be seen as the pump–probe time delay is scanned. This is a
Franck–Condon effect caused by oscillation of the wave
packet prepared by the 624 nm pulse on the potential energy
surface of the B state. This phenomenon has been exten-
By contrast, examine the transient produced from the
CH2I2 cell, shown in Fig. 4͑b͒. The pump pulse at 312 nm
causes the parent molecule to dissociate, producing I2 in the
D
state, which is fluorescent. At positive times ͑312 nm
Ј
pulse arriving first͒, the 624 nm ͑probe͒ pulse depletes the
population of the D state, causing a drop in signal intensity,
Ј
see Figs. 4͑b͒ and 5͑b͒. Unlike the case of the iodine sample,
however, the signal does not become zero when the 312 nm
pulse arrives first; it is simply at a lower level. At negative
times, the probe pulse is unable to deplete the D state, so
Ј
the signal is at a higher level than at positive times. Several
other differences are also apparent; the signal from the CH2I2
sample shows no modulation for negative time delays, but
periodic modulations in the depleted fluorescence can be
seen at positive time. It is clear from the comparisons above
that the I2 fluorescence signal being produced from the CH2I2
sample is a result of photoinduced molecular detachment of
I2 molecules and not background iodine in the cell.
Since the threshold for production of I2 from the photo-
dissociation of CH2I2 is 9.4 eV7,8 and the energy of a 312 nm
photon is 4 eV, multiphoton absorption is required to pro-
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:
128.143.23.241 On: Mon, 22 Dec 2014 20:33:30

J. Chem. Phys., Vol. 109, No. 11, 15 September 1998
Marvet et al.
4419
FIG. 6. Power dependence data for the photodetachment of I (D ) from
Ј
2
CH2I2. Data are plotted as fluorescence intensity at 340 versus intensity of
the 312 nm laser pulse. The least-squares fit displayed was obtained with a
dependence of the fluorescence intensity on 312 nm laser power of 3.4.
duce the observed fluorescence signal. Reliable power de-
pendence measurements for high order nonlinear processes
of this type are inherently difficult because of the necessity
for a wide dynamic detection range and because these types
of processes can easily become saturated. We have taken
care in measuring the power dependence to ensure that the
laser pulse does not saturate either the molecular transition or
the instrumental response even when it is at its most intense.
Power dependence data were collected by deliberately de-
feating the synchronization between the oscillator and ampli-
fier of the femtosecond laser; this produces a train of pulses
with a wide range of random intensities. Several hundred
points were then recorded by simultaneously detecting the
fluorescence signal from the product cell using the PMT and
the laser intensity using a fast photodiode, taking care not to
saturate either detector. Figure 6 shows the data plotted as
fluorescence intensity at 340 nm versus intensity of the 312
nm laser pulse. A least-squares fit of the data shows a nearly
cubic power dependence, indicating that the pump excitation
is a three-photon process.
FIG. 7. Time-resolved data from the multiphoton dissociation of alkyl ha-
lides at 312 followed by probing at 624 nm and detection of the D →A
Ј
Ј
transition. ͑a͒ From the dissociation of CH I , by detection of I (D ) at 340
Ј
2
2
2
nm. The inset shows the positive time data multiplied by a factor of ten,
along with a least-squares fit obtained using a bimodal Gaussian distribution
of vibrational levels ͑see text͒. The observed fit was obtained using two
modes, the first centered at v ϭ19 and having a FWHM of 7. The second
Ј
contribution is from a mode with an oscillation frequency of 104 cmϪ1
,
having a FWHM approximately equivalent to one vibrational level in the D
Ј
state. ͑b͒ Time-resolved data of the molecular detachment of Br2 from
CH2Br2 show depletion at positive times, and a strong time zero feature. The
transient was obtained by detecting the Br D →A fluorescence at 287 nm.
Ј
Ј
2
͑c͒ The transient of CH2Cl2 was obtained by detection at 254 nm. This
reaction yielded much lower signals, therefore the noise level is higher, but
depletion by the 624 nm pulse at positive times is clear.
Figure 7 also clearly shows vibrational coherence in the
I2 fragment after photodissociation of CH2I2. This is an in-
dication that the process responsible for formation of the
halogen molecules is concerted, since in-phase vibrations
could not result unless the molecules are formed within a
short time of each other and on the same region of the prod-
uct potential energy surface ͑PES͒. The fundamental fre-
Figure 7 shows time-resolved data obtained from the dis-
sociation of CH2I2, CH2Br2 and CH2Cl2. Each transient was
obtained by multiphoton excitation at 312 nm, followed by
quency of Br in the D state is 150.86 cmϪ1 32
which cor-
,
Ј
2
depletion probing at 624 nm from the D state of the relevant
responds to a vibrational period of 221 fs. While vibrational
coherence in the Br2 fragment has not to date been resolved,
further efforts to do so are underway. The fundamental fre-
Ј
diatomic halogen product. The polarization vectors of pump
and probe pulses were aligned parallel to each other. Fluo-
rescence was detected at the wavelength corresponding to the
maximum intensity of the appropriate D →A transition.
quency of Cl in the D state is 252.43 cmϪ1 34
which cor-
,
Ј
2
responds to a vibrational period of 132 fs. Since observation
of vibrational modulation depends upon the convolution of
the molecular dynamics with the pulses, a 60 fs pulsewidth
does not allow us to observe vibrational coherence in the Cl2
fragment. The low signal to noise ratio in both the spectral
and dynamic data from the CH2Cl2 sample is due to rather
weak fluorescence from the nascent Cl2 fragment.
Figure 8 shows transients obtained from the CH2I2 cell
when pump and probe lasers were polarized parallel and per-
pendicular to each other. There is not sufficient anisotropy in
Ј
Ј
Figure 7 shows that in each case, the fluorescence is depleted
at positive times ͑probe pulse following pump, to the right of
time zero͒. The intense features at time zero ͑pump and
probe pulses overlapped in time͒ are due to a cooperative
multiphoton effect which enhances the fluorescence signal.
In this region, the molecules absorb 312 and 624 nm photons
simultaneously. This leads to an enhancement in the fluores-
cence signal if it opens up another reaction pathway for the
production of X (D ).
Ј
2
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:
128.143.23.241 On: Mon, 22 Dec 2014 20:33:30

4420
J. Chem. Phys., Vol. 109, No. 11, 15 September 1998
Marvet et al.
volvement of Rydberg state excitation in the I2 photodetach-
ment process. Since molecular Rydberg states have very long
lifetimes, particularly at high excitation energies ͑from Ϸ100
fs for low N to ϳ␮s for high N͒, the time-resolved data
presented in this work also indicate that Rydberg states are
not involved in the molecular detachment process.
In our experiments we observed no evidence of molecu-
lar iodine products in any of the lower valence states. How-
ever, I2 has a dissociation energy of 1.54 in its ground state,
0.54 in its B state and Ϸ0.3 eV in its A and A states,39 so
Ј
molecular product in any of these states would dissociate
rapidly under the experimental conditions because the avail-
able energy is very high. An ion-pair state, which correlates
to XϩϩXϪ, would be a good candidate for a dissociation
product because these states are strongly bound and have
long range attractive forces. For halogens there are 18 of
3
1
1
these, corresponding to the P, D and S terms of the Xϩ
ion.
First tier ion-pair states have equilibrium energies within
0.16 eV of each other.40 The two other ion-pair families are
found approximately 0.9 and 1.5 eV higher.39 Three-photon
excitation at 312 nm is equivalent to 12 eV, which translates
into an excess energy of Ϸ3 eV above the observed barrier to
molecular photodetachment at 9.4 eV. This brings all the
ion-pair states within energetic reach. The observed data for
CH2I2 photodissociation indicate that only a small percent-
age ͑10%, not corrected for fluorescence yield or detection
efficiency͒ of fluorescence occurs at wavelengths between
250 and 290 nm. These wavelengths correspond to the sec-
ond tier of ion-pair states, which correlate with Xϩ(3P0)
FIG. 8. Time-resolved data of the CH2I2 dissociation obtained with parallel
and perpendicular configurations. Parallel and perpendicular refer to the
polarization vectors of the pump and probe pulses relative to each other.
Note the difference in intensity of the time zero feature between the parallel
and perpendicular polarization configurations. The anisotropy data are am-
biguous; no reliable conclusions about the rotational population of the prod-
ucts can be drawn from analysis of these data.
the data to obtain reliable information about the rotational
dephasing time of the molecules. The fact that time zero
enhancement is greater when the 624 nm pulse is polarized
parallel to the 312 nm pulse further confirms that this feature
is due to a cooperative process.
ϩX(1S).36 The observed predominance of the D state
Ј
strongly suggests that electronic excitation directly correlates
to this state. However, it is also possible that high-energy
excitation of the parent molecule yields a halogen molecule
in another ion-pair state while it is still close to the carbene
fragment, and that under these circumstances relaxation to
IV. DISCUSSION
A. Spectroscopy
the D state occurs. For example, the dissociation process
Ј
itself may act as a half-collision. It is known that single
collisions of most I2 ion-pair states with buffer gases are
The excited state of CH2I2 which is being accessed by
the excitation is unknown. The first excited state has an ab-
sorption near 32 000 cmϪ1, which corresponds to a single
photon at 312 nm. This state has B1 symmetry;10–12 for a
three-photon process, resonant enhancement via this state
would also reach a B1 dissociative state. However, a state of
this symmetry has a node between the iodine atoms, which
precludes the formation of an interhalogen bond. Two-
photon resonant enhancement at 312 nm allows access to the
B2 and A1 excited states; there are no symmetry restrictions
on forming a bond between the iodine atoms via transitions
having either symmetry.
In an effort to elucidate the nature of the parent elec-
tronic state that correlates to dihalogen products, Okabe
et al. compared the absorption spectrum of CH2I2 in the
vacuum ultraviolet region with the fluorescence excitation
spectrum in the same region ͑342 nm detection͒.7 The ab-
sorption and fluorescence excitation spectra both showed
highly efficient in causing nonradiative transitions to the D
Ј
state;33 it is possible that the half-collision represented by the
dissociation may be efficient enough to produce a similar
effect.
The fits to the I2 spectrum shown in Fig. 3͑a͒ indicate
that the population of the D state is non-Boltzmann, i.e.,
Ј
nonstatistical. This would be expected in a molecule which
had been excited using a femtosecond pulse, which produces
a wave packet on the excited potential energy surface. The
fact that the coherence in the parent molecule is preserved in
the dissociation fragment, however, indicates that the reac-
tion is completed very rapidly, before thermalization can oc-
cur. This is consistent with the observed reaction time and
the vibrational coherence in the product channel.
B. Transition state dynamics
*
broad continua, which were ascribed to C–I ␴→␴ transi-
tions. Although features assignable to Rydberg transitions
appeared in the absorption spectrum, their absence from the
fluorescence excitation spectrum seems to exclude the in-
Figures 7 and 8 both show that there is a large enhance-
ment at time zero in the 340 nm fluorescence signal from the
photoinduced detachment of halogen molecules from gem–
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:
128.143.23.241 On: Mon, 22 Dec 2014 20:33:30

J. Chem. Phys., Vol. 109, No. 11, 15 September 1998
Marvet et al.
4421
FIG. 9. Possible excitation schemes to produce time
zero fluorescence enhancement. The threshold indicated
is the observed excitation threshold for production of
CH2 and I (D ). The upper line on the figure corre-
Ј
2
sponds to 12 eV, the excitation energy corresponding to
a three-photon pump transition. ͑a͒ One-photon excita-
tion with the 312 nm ͑pump͒ pulse requires three pho-
tons of 624 nm ͑probe͒ light to reach the threshold for
production of I (D ). An additional photon of the
Ј
2
probe produces an excitation of 12 eV. Single-photon
absorption at 312 nm is equivalent to an excitation of 4
1
˜
eV, which is sufficient to reach the A( B1) state of
CH2I2 ͓marked ͑i͒ in the figure͔. ͑b͒ An initial absorp-
tion of two pump photons corresponds to an excitation
of 8 eV, requiring one photon of 624 nm light to reach
the threshold and two to be equivalent to an excitation
of 12 eV. The pump-excited state, marked ͑ii͒, is un-
known. ͑c͒ A three-photon pump corresponds to an ini-
tial excitation of 12 eV. This is the pathway of the most
interest to us because the dynamics observed between
the pump and probe pulses corresponds to the transition
state of the photoinduced molecular detachment reac-
tion. The state ͑iii͒ is the dissociative state under inves-
tigation.
dihaloalkanes. This is of interest both because it can be used
to yield information about the electronic structure of the par-
ent molecule and because it affords a measure of the lifetime
of the transition state for the reaction. Analysis of this can
help to elucidate the nature of the reaction mechanism. We
need to consider possible causes of the observed enhance-
ment.
Figure 9 shows a schematic of the possible excitation
schemes that can lead to enhancement of the fluorescence
signal at time zero. The decay time of the time zero fluores-
cent enhancement is directly related to the lifetime of the
initial excited state of the parent molecule accessed by the
pump transition ͑which we will call the pump-excited state͒.
The molecular dynamics is manifested as an asymmetry in
the time zero peak; the decay time ͑on the positive side of
time zero͒ is slower than the rise time. This allows us to
determine an upper limit for the lifetime of the pump-excited
state. It should be noted that the pump–probe data at time
zero measure the statistical lifetime of the pump-excited
state, not the time taken to reach a particular region of the
potential energy surface. These data thus contain no direct
information about the exit channel dynamics, but can be used
to measure an upper limit for the dissociation time of the
molecules.
The transition state of the photoinduced molecular de-
tachment reaction under investigation is the state of the par-
ent molecule accessed by a three-photon pump transition; an
excitation scheme for this process is shown in Fig. 9͑c͒.
However, other excitation processes are also possible at time
zero. In order to have the observed effect, the process re-
sponsible for the fluorescence enhancement must be of the
general type shown, i.e., one or more 312 nm ͑pump͒ pho-
tons causes a transition to an excited electronic state. Subse-
quent absorption of 624 nm ͑probe͒ photons produces CH2X2
sorption of one 312 nm photon, it requires three photons of
624 nm light to reach 10 eV, above the observed threshold
for production of I in the D state,7 four ͑plus the initial one
Ј
2
photon at 312 nm͒ provides the same energy as three photons
of 312 nm light. This is shown as scheme ͑a͒. In either case,
the state being probed is 4 eV above the ground state; ob-
served molecular dynamics would give the statistical lifetime
of this state. On the other hand, Fig. 9͑b͒ shows the situation
if the pump-excited state is reached by two pump photons.
There are again two possibilities; a one-photon probe transi-
tion will exceed the threshold for production of I (D ); a
Ј
2
two-photon probe transition accesses the same state that is
reached by three photons at 312 nm. If this were the excita-
tion scheme, the lifetime of the state 8 eV above the ground
state would be reflected in the molecular dynamics. The third
scheme is shown in Fig. 9͑c͒. In this case, three photons of
ultraviolet ͑UV͒ are absorbed, thus reaching the state of
CH2I2 that we wish to examine. The lifetime of this state is
probed by excitation by the 624 nm pulse to a higher energy
state.
We have observed a significant enhancement to the fluo-
rescence at time zero even for quite low intensities of 624
nm pulses. This implies that the scheme shown in Fig. 9͑a͒ is
probably not a likely alternative, since it requires at least
three 2 eV photons to be absorbed as the probe. Scheme ͑c͒
would allow the lifetime of the transition state to be deter-
mined; scheme ͑b͒ would yield molecular dynamic informa-
tion characteristic of the lifetime of another electronic state
than the one of interest. We cannot determine at this time
which is more likely. However, it should be noted that be-
cause we are measuring an upper limit for the dissociation
time, the process that produces the depletion signal cannot be
slower than the dissociation time measured using the transi-
tion state dynamics.
in a sufficiently high energy state to produce CH ϩX (D ).
In order to analyze the transition state dynamics of the
reaction, a kinetic model was constructed for the behavior of
the fluorescence signal as a function of delay time between
the pump and probe pulses. Figure 10 illustrates the model.
Ј
2
2
There are a number of possibilities for this excitation
scheme; the most plausible are shown in Fig. 9 for the ex-
ample of CH2I2. If the initial excitation is provided by ab-
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:
128.143.23.241 On: Mon, 22 Dec 2014 20:33:30

4422
J. Chem. Phys., Vol. 109, No. 11, 15 September 1998
Marvet et al.
FIG. 11. Time-resolved data of the molecular photodetachment of I2 from
CH2I2 and n-C4H8I2. The data are plotted as points; the fit is shown as a line
in each case. The difference in dissociation time between the two molecules
can be accounted for completely by the difference in mass of the alkyl
fragment. The relatively poor fit to the C4H8I2 dynamics at positive times is
due to the presence of a vibrational oscillation at early times.
FIG. 10. Kinetic model for the dissociation of CH2I2. The model assumes
that the signal is comprised of two contributions: ͑a͒ Depletion signal. This
is produced by the molecular detachment process. Excitation of CH2I2 with
three photons of 312 nm light produces an excited molecule, which disso-
the temporal width of the pulses, this function is convoluted
with a Gaussian; the resulting curve is shown in Fig. 10͑a͒.
To model the intense time zero feature, a similar ap-
proach was used. In this case, there is no signal at negative
times. At time zero a cooperative ͑multiphoton͒ effect occurs
and signal levels reach their maximum value. As the time
delay between pump and probe pulses increases, the mol-
ecules begin to dissociate and the multiphoton signal en-
hancement is no longer possible. Thus the time zero feature
can be represented as a half exponential function as shown in
Fig. 10͑b͒. Again, this is convoluted with a Gaussian to ac-
count for the temporal pulsewidth. The overall signal should
then behave as a sum of contributions from the depletion and
time zero features as shown in Fig. 10͑c͒. For simplicity,
variations in the signal caused by rotational and vibrational
motion of the nascent products were not included. The model
does not take into account the possibility of contributions
from other states, but should provide a reliable upper limit on
the dissociation time for a given pulsewidth. The details of
the mathematical formulation used are included in the Ap-
pendix.
ciates into CH and I (D ). Depletion of the D fluorescence with a 624 nm
Ј
Ј
2
2
pulse probes the dynamics of formation of I (D ). ͑b͒ Time zero signal.
Ј
2
When the pump and probe pulses coincide, it is possible for CH2I2 to absorb
probe photons prior to dissociation, which produces a highly excited parent
molecule. This increases the amount of D →A fluorescence observed be-
Ј
Ј
cause it opens another pathway for production of I (D ). In this case, the
Ј
2
statistical lifetime ␶ of the pump-excited state determines for how long after
the initial excitation the molecule can absorb the probe. The overall signal is
a weighted sum of these two contributions, each of which is convoluted with
a Gaussian to simulate the pulsewidth.
The observed transient can be modeled as the sum of two
contributions. The molecular detachment signal ͑a͒ will ex-
hibit a step at time zero due to depletion by the 624 nm
pulse. The time zero fluorescent enhancement ͑b͒ can be rep-
resented as a Gaussian-type peak decaying exponentially at
positive time.
Figure 10͑a͒ illustrates the depletion signal. Three pho-
tons of 312 nm light are absorbed by CH2I2; the molecule
dissociates to produce I in the D electronic state. Detection
Ј
2
of fluorescence from the D →A transition at 340 nm allows
Ј
Ј
reaction dynamics to be probed by monitoring the amount of
D state population depleted by absorption of a second fem-
Transients obtained near time zero from methylene diio-
dide and butylene diiodide are shown in Fig. 11. Both were
taken with the polarization vectors of pump and probe lasers
aligned parallel to each other. The data were fit using a non-
linear least-squares procedure and the model described
above. The dissociation time of C3H7CHI2 was found to be
␶ р87Ϯ5 fs, indicating that molecular detachment is
prompt. Fitting of the CH2I2 time zero transient using the
same method produced a dissociation time ␶ of 47Ϯ3 fs. If
we assume that the difference in dissociation time of the two
molecules is caused only by the difference in reduced mass
Ј
tosecond pulse at 624 nm. When the probe pulse arrives
before the 312 nm pulse ͑negative time͒, it does not affect
the intensity of the fluorescence. Since it is the nascent prod-
uct that is being probed, depletion starts to become possible
once the pulses are overlapped in time; at this point, the
signal level begins to drop. The amount of depletion possible
increases steadily as the molecules dissociate; this continues
until no more I (D ) can be formed. The molecular signal
Ј
2
response as a function of pump–probe time delay would then
be a step function with an exponential decay. To account for
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:
128.143.23.241 On: Mon, 22 Dec 2014 20:33:30

J. Chem. Phys., Vol. 109, No. 11, 15 September 1998
Marvet et al.
4423
TABLE I. Vibrational frequencies of CH2I2 normal modes in the ground state. Presented are the calculated ͑gas
phase͒ and experimentally measured solution and liquid phase values. The vibrational frequencies were ob-
tained at the Hartree–Fock ͑HF͒ level by optimizing the geometry of the molecule and then conducting a
frequency analysis using GAUSSIAN 94 ͑Ref. 42͒. The LANL2DZ effective core potential basis set was used for
*
the iodine atoms and the 6-31G basis set for the hydrogen and carbon atoms. The calculated frequencies of the
normal modes were scaled by 0.8929 and the zero-point energy by 0.9135 ͑Refs. 43 and 44͒.a
Experimental
Calculated
Gas phaseb
Liquid CH2I2
͑Raman͒
Liquid CH2I2
͑Infrared͒
Cyclohexane
solution ͑Raman͒
c
c
d
Peak
Type
␯
␯
␯
␯
I-C-I bend
117
454
552
121
486
567
134
493
581
4
I-C-I symmetric stretch
I-C-I asymmetric stretch
CH2 rock
486
573
717
3
9
716
714
714
7
␯
␯
␯
␯
Torsion
CI2 rock
H-C-H bend
H–C–H symmetric stretch
H–C–H asymmetric stretch
1056
1148
1374
2958
3043
1028
1104
1350
2968
3049
1028
5
1106
1351
2967
3049
8
1369
2968
3049
2
1
␯
6
aAll values given in cmϪ1
.
bThis work; calculated using GAUSSIAN 94 as stated in text.
cSee Reference 45.
dReference 18.
C. Vibrational analysis
␮, we can use the following relationship to predict the ex-
pected difference in dissociation time:
The transient shown in Fig. 7 was produced from the
dissociation of CH2I2 when pump and probe lasers were po-
larized parallel to each other. The modulation in signal in-
tensity at positive times is indicative of vibrational coherence
in the I2 product, which requires that the molecular detach-
ment process be concerted. It was not possible to perform
Fourier transform analysis of the data because the vibrational
modulation dephases quickly and does not seem to recur ͑see
Fig. 8͒. An alternative method had to be found, therefore.
The simplest method for analysis of a rapidly dephasing os-
cillatory signal of this type is a damped sinusoidal function
of the form,
␶
␮
␮
MeI
BuI
2
2
ϭ
.
͑1͒
ͱ
ͩ ͪ
␶
BuI
2
MeI
2
This expression applies if the fragments have the same
amount of kinetic energy in both cases, the right-hand side
yields a ratio of 0.538. The observed ratio obtained from the
fit ␶
/␶ 2ϭ0.54. Thus the near factor of two difference
BuI
MeI
2
in dissociation time can be completely accounted for by the
mass difference, i.e., there appears to be no IVR occurring
during dissociation even for a system like gem–diiodobutane
in which there is a moderately large density of states.
To estimate the amount of available energy that is parti-
tioned into center-of-mass translational motion of the frag-
ments, we use ␶ as an upper limit for the dissociation time of
each molecule. For the purposes of the model, we will as-
sume ␶ to be the time taken for the fragments to move a
distance L apart, at which point dissociation will be consid-
ered to be complete. The amplitude of the C–I stretch is Ϸ1
Å ͑see Table I͒, so it is reasonable to assume that the
carbon–iodine bonds can be considered broken once the
fragments have moved more than 2 Å farther apart than the
equilibrium distance. A simple impulsive model can then be
employed to determine the translational energy E of the frag-
ments by assuming that they reach terminal velocity instan-
taneously,
AϩBeϪt/␶ cos ␻tϩ␾͒,
͑3͒
͑
where A and B are experimentally determined parameters, ␶
is the dephasing time, ␻ the vibrational frequency and ␾ the
phase. The data shown in Fig. 7 were fit using this model and
were found to have an oscillation frequency of 92.2 cmϪ1
,
corresponding to a vibrational period of 362 fs. The dephas-
ing time was ϳ1 ps.
Time-resolved dynamics collected at 285 and 272 nm
have also been shown to exhibit vibrational coherence9; if
this is indeed due to population in the f state ͑vide supra͒, the
f(3⌸ϩ0g)→B(3⌸0u) fluorescence would be expected to con-
tribute to the coherences observed in the 340 nm region. The
observed dephasing would then be due in part to this addi-
tional contribution to the time-resolved signal. To test this
supposition, the vibrational dynamics shown in Fig. 7͑a͒
were fit using a bimodal-type model. The displayed fit was
obtained by assuming a bimodal Gaussian distribution of vi-
2
L
Eϭ 1␮
,
͑2͒
ͩ ͪ
2
␶
where ␮ is the reduced mass of the carbene and I2 moieties,
assuming they form a pseudodiatomic ͑essentially the mass
of the alkyl fragment͒. Using a length parameter of L
ϭ2.0 Å and a dissociation time ␶ of 47 fs, the kinetic energy
of CH2I2 upon dissociation is estimated to be 1.26 eV.
brational levels. One mode is centered at ϭ19 in the D
v
Ј
Ј
state with a full width at half maximum ͑FWHM͒ of 7. This
corresponds to a vibrational spacing of 96.6 cmϪ1, which
reflects an oscillation period of 345 fs and is close to the
This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP:
128.143.23.241 On: Mon, 22 Dec 2014 20:33:30

4424
J. Chem. Phys., Vol. 109, No. 11, 15 September 1998
Marvet et al.
FIG. 13. Normal mode analysis of the ground state of CH2I2, plotted as a
function of I–I and CH2–I distance. Only the I–C–I bending (␯4) and
I–C–I symmetric stretch (␯3) contribute to the I–I vibration. The other
normal modes have limited or no motion between the iodine atoms.
FIG. 12. Time-resolved data obtained near time zero for the high-energy
dissociation of CH2I2 and CD2I2. The dissociation time is identical for the
two compounds, indicating that vibrational modes in the alkyl fragment are
not involved in the dissociation mechanism ͑see text͒.
result from the damped sinusoidal model discussed above.
The second mode has an oscillation frequency of 104 cmϪ1
ous vibrational modes of the parent molecule to the molecu-
lar detachment process, a frequency analysis of the ground-
state equilibrium geometry of CH2I2 was performed using
and a FWHM equivalent to a single D vibrational level.
Ј
42
This mode appears to be contributing approximately 15% of
the time-dependent signal at 340 nm, which is consistent
with the intensity of the f→B fluorescence expected in this
region.
Least-squares fitting to the dispersed fluorescence spec-
trum in the 290–350 nm region yielded a distribution cen-
GAUSSIAN 94 ͑for technical details see Table I caption͒.43,44
Table I compares the normal mode frequencies of the
calculated gas phase values to the experimental liquid45 and
solution phase values.18 Despite the fact that the calculated
and experimental values are determined for different phases,
it can be seen that the calculated frequencies are close to the
experimentally observed ones ͑0.1%–9% deviation͒. The
larger errors occur for those modes involving mostly iodine
atom motion. This observation is not unexpected because ab
initio calculations on molecules containing many-electron at-
oms like iodine are inherently less precise.
To calculate the extent of motion along the reaction co-
ordinate for each normal mode the displacements of each
atom in each normal mode were added to the equilibrium
geometry configuration and single point HF calculations
were conducted on these modified geometries. The energy
from each single point calculation was added to the scaled
zero-point energy and the resulting energy was divided by
the scaled frequency for the normal mode. The result is the
number of quanta in that normal mode corresponding to the
atom positions calculated from the displacements. Then the
normal mode displacements were scaled to find the atom
positions that would correspond to one quantum of energy in
each normal mode. Finally, these modified atom positions
for each normal mode were mapped to the motion of the
molecule along two coordinates: the distance between the
two iodine atoms ͑I...I͒ and the distance between the CH2
fragment and the I2 fragment (H2C...I2) ͑measured from the
carbon atom to the midpoint of the iodine atoms͒.
tered at
ϭ9 of the D state; see Figs. 3͑a͒ and 3͑b͒.
Ј
v
Ј
Clearly, there is a discrepancy between this result and the
value obtained from fitting the dynamic data. The reason for
this is not clear but may be due in part to the presence of
fluorescence at 340 nm originating from the f→B transition.
It is clear from comparison with the fit to the spectrum
of I2 in Ar ͑Fig. 3͒ that the vibrational population is not
thermal; both the spectrum and the vibrational oscillations
from the CH2I2 sample can be fit by a Gaussian distribution
of populations in the vibrational levels of the D state.
Ј
Femtosecond studies of the reaction dynamics of Ryd-
berg states of methyl iodide showed a substantial difference
in reaction time for the photodissociation of CH3I and
CD3I.41 The explanation proposed for this observation was
that vibrational modes in the alkyl fragment are involved in
the predissociation process, thus making the isotope effect
for the dissociation substantial. Figure 12 shows the femto-
second transients near time zero for CH2I2 and CD2I2; it is
apparent from the data that there is no detectable difference
in dissociation time between the two molecules. This is
strong evidence that vibrational modes in the alkyl fragment
are not involved in the dissociation mechanism.
In order to investigate the possible contribution of vari-
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:
128.143.23.241 On: Mon, 22 Dec 2014 20:33:30

J. Chem. Phys., Vol. 109, No. 11, 15 September 1998
Marvet et al.
4425
Figure 13 shows the results of the normal mode analysis,
plotted as a function of I–I and CH2–I2 distance. For each
mode the amplitude of the motion corresponds to one quan-
tum of excitation in that mode. It is apparent that only the
I–C–I bending ␯ and I–C–I symmetric stretch ␯ contrib-
ute to the I–I interatomic distance. The other normal mode
has limited or no motion between the iodine atoms. The I–
C–I bending contributes significantly to the H2C...I2 coordi-
which the charges are mostly concentrated on the iodine at-
oms; there is little redistribution of charge on the carbon
atom.23 This is consistent with a charge-transfer mechanism
of the type proposed, even though the concerted elimination
of interest in this work proceeds from a highly excited elec-
tronic state.
4
3
V. CONCLUSIONS
nate. Because ␯ is the lowest energy mode, it is expected to
4
Photoinduced molecular detachment processes have
been investigated by femtosecond pump–probe spectroscopy
for methylene iodide and some related haloalkanes. It was
found that multiphoton excitation of these molecules at 312
nm gives rise in every case to halogen molecules in ion-pair
be significantly populated at room temperature; at 294 K
approximately 10% of the molecules have four quanta in this
mode. The effects of vibrational excitation in the parent on
the molecular detachment process are currently under inves-
tigation.
states, predominantly the D state. Femtosecond time-
Ј
resolved experiments performed on these dihaloalkanes
show that the molecular detachment processes are extremely
fast ͑Ͻ50 fs͒. Due to the large enhancement in the 340 nm
fluorescence at time zero feature, it was not possible to un-
ambiguously observe rotational anisotropy in the data.
Time-resolved data for the dissociation of methylene io-
dide revealed the presence of vibrational coherence in the
halogen product. Analysis of the vibrational frequency of the
D. Mechanism
Photodissociation studies of alkyl iodides at 304 nm ex-
citation energy revealed that the partitioning of available en-
ergy into IVR increased from 19% for methyl iodide to 70%
in n-butyl iodide, an energy difference of 1 eV.46,47 This
contrasts directly with our results, which indicate no appar-
ent increase in IVR in the alkyl fragment on going from
diiodomethane to gem–diiodobutane. A direct recoil with
C2v symmetry allows relatively efficient coupling between
translational motion of the ␣-carbon and internal vibrational
modes of the alkyl fragment.22 In this case, we would expect
to see an increase in IVR on dissociation of gem–
diiodobutane as compared to methylene iodide. Based on the
dissociation time evidence, it therefore appears that the mo-
lecular detachment of halogen molecules from dihaloalkanes
signal confirmed the electronic state assignment of the D
Ј
state. The fit to the phase of the oscillations indicates that the
two iodine atoms are at a turning point of their vibration at
time zero. The observation of vibrational coherence indicates
that the photodissociation is concerted, i.e., processes ͑b͒ or
͑c͒ in Fig. 1 are the most likely mechanisms for this reaction.
Mechanism ͑a͒ can be ruled out as a source of the observed
I2 dynamics because if this were the pathway it would not
have been possible to observe vibrational motion by the
pump–probe method described.
¨
proceeds by a different pathway. A Huckel frontier molecu-
lar orbital study of dissociation at lower energies indicated
that the ‘‘least-motion’’ path ͑in which C2v symmetry is re-
tained͒ is a high-energy pathway and that when the carbene
moiety is allowed to slip off-center from the halogen–
halogen axis, the reaction path is considerably stabilized.22
This corresponds to a pathway for the concerted elimination
of the general type shown in Fig. 1͑c͒. We tentatively ad-
vance this pathway as the mechanism for the molecular pho-
todetachment process.
Because the carbene radical is ambiphilic, i.e., the lone
pair at the front of the radical behaves as a nucleophile and
the empty p␲ orbital is electrophilic, an asynchronous con-
certed mechanism as discussed above would tend to produce
charge separation in the halogen fragment.22 This is consis-
tent with the fact that only ion-pair states of halogen mol-
ecules have been observed on the photodetachment of halo-
gens from dihaloalkanes. In this case, an ylide-type
intermediate would be stabilized by the charge distribution
of the carbene fragment. Iso-diiodomethane, H2–C–I–I has
been observed in gas matrices at low energies;48 in calcula-
tions of this system, it was found that less than 2 eV of
excitation energy is required to produce this species. This is
less energy than is needed to form CH2IϩI.23 However, the
bonding between the iodine atoms in this species is signifi-
cantly weaker than in I2 or IϪ2 , which may explain why
CH2ϩ2I is the more common reaction pathway. The iodine
atoms in this molecule are described as a contact ion pair in
ACKNOWLEDGMENTS
We thank Dr. P. Gross for writing the routines used for
fitting spectra and for insightful discussions. We are grateful
to Professor J. F. Harrison at MSU for his help with the ab
initio calculations. The butyl diiodide used in these experi-
ments was synthesized by Lauren Heystek. This work was
partially funded by a Camille and Henry Dreyfus New Fac-
ulty Award. U.M. acknowledges funding from a James L.
Dye Endowment Fellowship. E.J.B. is an NSF Doctoral Fel-
low. M.D. is a Beckman Young Investigator and a Packard
Science and Engineering Fellow.
APPENDIX
In order to simulate the transition state dynamics of
the molecular detachment reaction, the signal is treated as
a sum of two contributions so that Itotal(t)ϭIdepletion(t)
ϩItimezero(t), as shown in Fig. 10.
To model the depletion dynamics, we will use a function
of the form shown in Fig. 10͑a͒. This can be represented
mathematically by ␩(t) eϪt/␶Ϫ1 , where ␩(t) is a step
͓
͔
function and ␶ represents the decay time of the transition
state, which we will assume decays as an exponential. This
function will be convoluted with a Gaussian to simulate the
effects of the temporal width of the pulse,
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:
128.143.23.241 On: Mon, 22 Dec 2014 20:33:30

4426
J. Chem. Phys., Vol. 109, No. 11, 15 September 1998
Marvet et al.
1 H. Okabe, Photochemistry of Small Molecules ͑Wiley, New York, 1978͒.
2 H. Okabe, A. H. Laufer, and J. J. Ball, J. Chem. Phys. 55, 373 ͑1971͒.
3 L. J. Butler, E. J. Hintsa, S. F. Shane, and Y. T. Lee, J. Chem. Phys. 86,
2051 ͑1987͒.
1
t2
x2
t
exp Ϫ
ꢀ ␩ t͒ exp Ϫ Ϫ1
͑
ͩ
ͪ
ͫ
ͩ ͪ
ͬ
␶
xͱ␲
4 P. J. Dyne and D. W. G. Style, J. Chem. Soc. 1952, 2122.
5 D. W. G. Style and J. C. Ward, J. Chem. Soc. 1952, 2125.
6 C. Fotakis, M. Martin, and R. J. Donovan, J. Chem. Soc., Faraday Trans.
2 178, 1363 ͑1982͒.
2
ϱ
1
tϪ␦͒
t
͑
ϭ
exp Ϫ
exp Ϫ dt
͵
ͩ
ͪ ͩ ͪ
x2
␶
xͱ␲
0
2
ϱ
7 H. Okabe, M. Kawasaki, and Y. Tanaka, J. Chem. Phys. 73, 6162 ͑1980͒.
8 G. Black, Research on High Energy Storage for Laser Amplifiers ͑Stan-
ford Research Institute, 1976͒.
1
tϪ␦͒
͑
Ϫ
exp Ϫ
dt,
͑A1͒
͵
ͩ
ͪ
x2
xͱ␲
0
9 Q. Zhang, U. Marvet, and M. Dantus, J. Chem. Phys. 109, 4428 ͑1998͒,
following paper.
where xϭ␴/
ͱ
(4 ln 2) and ␴ is the FWHM of the pulse. In-
10 M. Kawasaki, S. J. Lee, and R. Bersohn, J. Chem. Phys. 63, 809 ͑1975͒.
11 P. M. Kroger, P. C. Demou, and S. J. Riley, J. Chem. Phys. 65, 1823
͑1976͒.
tegrating this expression over t produces the dynamic model
for depletion of the molecules,
12 S. L. Baughcum and S. R. Leone, J. Chem. Phys. 72, 6531 ͑1980͒.
13 W. H. Pence, S. L. Baughcum, and S. R. Leone, J. Phys. Chem. 85, 3844
͑1981͒.
2
1
x
␦
␦
x
␦
x
exp
Ϫ
1ϩerf
Ϫ
sign
Ϫ
ͯ ͯ
ͫ
ͩ ͪ ͬͫ
ͩ
ͪ
ͬ
ͩ
ͪ
2x
2␶
␶
x
2␶
x
2␶
14 T. F. Hunter and K. S. Kristjansson, Chem. Phys. Lett. 90, 35 ͑1982͒.
15 J. Zhang, E. J. Heller, D. Huber, D. G. Imre, and D. Tannor, J. Chem.
Phys. 89, 3602 ͑1988͒.
1
␦
␦
16 J. Zhang and D. G. Imre, J. Chem. Phys. 89, 309 ͑1988͒.
17 W. M. Kwok and D. L. Phillips, J. Chem. Phys. 104, 2529 ͑1996͒.
18 W. M. Kwok and D. L. Phillips, Chem. Phys. Lett. 235, 260 ͑1995͒.
19 B. J. Schwartz, J. C. King, J. Z. Zhang, and C. B. Harris, Chem. Phys.
Lett. 203, 503 ͑1993͒.
Ϫ
1ϩsign
erf
,
͑A2͒
ͯ ͯ
ͫ
ͩ ͪ ͩ ͪ
ͬ
2x
x
x
where ␦ indicates the time delay between the pump and
probe pulses.
20 K. Saitow, Y. Naitoh, K. Tominaga, and K. Yoshihara, Chem. Phys. Lett.
262, 621 ͑1996͒.
In order to model time zero, we need the convolution of
a delta function with a Gaussian. This is the second term in
Eq. ͑A1͒. Thus the complete solution for the molecular dy-
namics of the molecules at time zero can be represented as
shown,
21 F. Duschek, M. Schmitt, P. Vogt, A. Materny, and W. Kiefer, J. Raman
Spectrosc. 28, 445 ͑1997͒.
22 S. R. Cain, R. Hoffmann, and E. R. Grant, J. Phys. Chem. 85, 4046
͑1981͒.
23 M. N. Glukhovtsev and R. D. Bach, Chem. Phys. Lett. 269, 145 ͑1997͒.
24 Z. Kisiel, L. Pszczolkowski, W. Caminati, and P. G. Favero, J. Chem.
Phys. 105, 1778 ͑1996͒.
1
t2
x2
t
A
exp Ϫ
ꢀ ␩ t͒ exp Ϫ Ϫ1
͑
25 U. Marvet and M. Dantus, Chem. Phys. Lett. 256, 57 ͑1996͒.
26 U. Marvet, Q. Zhang, and M. Dantus, J. Phys. Chem. A 102, 4111 ͑1998͒.
27 Q. Zhang, U. Marvet, and M. Dantus, Faraday Discuss. 108, 63 ͑1997͒.
28 I. Pastirk, E. J. Brown, Q. Zhang, and M. Dantus, J. Chem. Phys. 108,
4375 ͑1998͒.
ͩ
ͪ
ͫ
ͩ ͪ
ͬ
␶
xͱ␲
1
t2
x2
t
ϩB
exp Ϫ
ꢀ ␩ t͒exp Ϫ ϩC
͑
ͩ
ͪ
ͩ ͪ
29 X. Zheng, S. Fei, M. C. Heaven, and J. Tellinghuisen, J. Chem. Phys. 96,
4877 ͑1992͒.
␶
xͱ␲
ϭA exp ␣2Ϫ␤͒ 1ϩsign ␥Ϫ␣͒erf
30 J. Tellinghuisen, J. Mol. Spectrosc. 94, 231 ͑1982͒.
31 H. Hemmati and G. J. Collins, Chem. Phys. Lett. 75, 488 ͑1980͒.
32 A. Sur and J. Tellinghuisen, J. Mol. Spectrosc. 88, 323 ͑1981͒.
33 J. B. Koffend, A. M. Sibai, and R. Bacis, J. Phys. ͑Paris͒ 43, 1639 ͑1982͒.
34 P. C. Tellinghuisen, B. Guo, D. K. Chakraborty, and J. Tellinghuisen, J.
Mol. Spectrosc. 128, 268 ͑1988͒.
͉
␥Ϫ␣͉
͒
͔
͑
͓
͑
͑
ϩBsign ␥͒erf
͉␥͉
͒ϩC,
͑A3͒
͑
͑
where ␣ϭx/2␶, ␤ϭ␦/␶ and ␥ϭ␦/x.
A, B and C are experimentally determined fitting param-
eters, since we cannot predict a priori the amount that each
component will contribute to the overall signal. We have
made a number of assumptions and approximations in this
model. First of all, in the interest of simplicity, the changes
in signal caused by rotational dephasing and vibrational co-
herence have not been accounted for. Also, the intensity pro-
file of the pulse in the depletion model is in fact a convolu-
tion of the three 312 nm pulses causing the initial excitation
with the 624 nm ͑probe͒ pulse. For the model, the temporal
width of the probe pulse as measured by frequency-resolved
optical gating ͑FROG͒ was used. This introduces little error
because the three-photon excitation will have a considerably
narrower temporal width than the 624 nm pulse; the cross-
correlation will therefore have a width very similar to that of
the probe. We have also assumed that the convoluted
FWHM of the pulses producing the time zero feature is the
same as that used for the depletion; the assumption is neces-
sary because the exact excitation process at time zero is not
known.
35 P. J. Jewsbury, K. P. Lawley, T. Ridley, F. F. Al-Adel, P. R. R.
Langridge-Smith, and R. J. Donovan, Chem. Phys. 151, 103 ͑1991͒.
36 K. Lawley, P. Jewsbury, T. Ridley, P. Langridge-Smith, and R. Donovan,
Mol. Phys. 75, 811 ͑1992͒.
37 P. J. Wilson, T. Ridley, K. P. Lawley, and R. J. Donovan, Chem. Phys.
182, 325 ͑1994͒.
38 M. Dantus, R. M. Bowman, and A. H. Zewail, Nature ͑London͒ 343, 737
͑1990͒.
39 R. S. Mulliken, J. Chem. Phys. 55, 288 ͑1971͒.
40 K. Wieland, J. B. Tellinghuisen, and A. Nobs, J. Mol. Spectrosc. 41, 69
͑1972͒.
41 M. H. M. Janssen, M. Dantus, H. Guo, and A. H. Zewail, Chem. Phys.
Lett. 214, 281 ͑1993͒.
42
GAUSSIAN 94 ͑Revision D.3͒, M. J. Frisch, G. W. Trucks, H. B. Schlegel, P.
M. W. Gill, B. G. Johnson, M. A. Robb, J. R. Cheeseman, T. Keith, G. A.
Petersson, J. A. Montgomery, K. Raghavachari, M. A. Al-Laham, V. G.
Zakrzewski, J. V. Ortiz, J. B. Foresman, J. Cioslowski, B. B. Stefanov, A.
Nanayakkara, M. Challacombe, C. Y. Peng, P. Y. Ayala, W. Chen, M. W.
Wong, J. L. Andres, E. S. Replogle, R. Gomperts, R. L. Martin, D. J. Fox,
J. S. Binkley, D. J. Defrees, J. Baker, J. P. Stewart, M. Head-Gordon, C.
Gonzalez, and J. A. Pople ͑Gaussian, Inc., Pittsburgh, PA, 1995͒.
43 A. P. Scott and L. Radom, J. Phys. Chem. 100, 16502 ͑1996͒.
44 J. A. Pople, A. P. Scott, M. W. Wong, and L. Radom, Isr. J. Chem. 33,
345 ͑1993͒.
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:
128.143.23.241 On: Mon, 22 Dec 2014 20:33:30

J. Chem. Phys., Vol. 109, No. 11, 15 September 1998
Marvet et al.
4427
45 F. L. Voelz, F. F. Cleveland, and A. G. Meister, J. Opt. Soc. Am. 43, 1061
47 W. K. Kang, K. W. Jung, D. C. Kim, K.-H. Jung, and H.-S. Im, Chem.
͑1953͒.
Phys. 196, 363 ͑1995͒.
46 W. K. Kang, K. W. Jung, K.-H. Jung, and H. J. Hwang, J. Phys. Chem.
48 G. Maier and H. P. Reisenauer, Angew. Chem. Int. Ed. Engl. 25, 819
͑1986͒.
98, 1525 ͑1994͒.
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:
128.143.23.241 On: Mon, 22 Dec 2014 20:33:30