Download as pdf or txt
Download as pdf or txt
You are on page 1of 22

Home Search Collections Journals About Contact us My IOPscience

Molecular photodissociation studied by VUV and soft x-ray radiation

This content has been downloaded from IOPscience. Please scroll down to see the full text.

2005 J. Phys. B: At. Mol. Opt. Phys. 38 S839

(http://iopscience.iop.org/0953-4075/38/9/025)

View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address: 137.132.123.69
This content was downloaded on 15/06/2014 at 07:26

Please note that terms and conditions apply.


INSTITUTE OF PHYSICS PUBLISHING JOURNAL OF PHYSICS B: ATOMIC, MOLECULAR AND OPTICAL PHYSICS

J. Phys. B: At. Mol. Opt. Phys. 38 (2005) S839–S859 doi:10.1088/0953-4075/38/9/025

Molecular photodissociation studied by VUV and soft


x-ray radiation
Kiyoshi Ueda1 and John H D Eland2
1 Institute of Multidisciplinary Research for Advanced Materials, Tohoku University,

Sendai 980-8577, Japan


2 Physical and Theoretical Chemistry Laboratory, Oxford University, South Parks Road,

Oxford OX1 3QZ, UK

E-mail: ueda@tagen.tohoku.ac.jp

Received 11 November 2004


Published 25 April 2005
Online at stacks.iop.org/JPhysB/38/S839

Abstract
A brief account of the developments in photodissociation studies of free
molecules by VUV and soft x-ray sources is given as an extended
introduction. Then two typical experimental setups are described for multiple-
ion coincidence momentum imaging and high-resolution Auger electron–ion
coincidence momentum imaging. Finally, some individual cases of molecular
dissociation following core-hole creation are examined, to illustrate how the
complex multidimensional data produced can be represented and interpreted.
The new experimental techniques based on Coulomb explosion are shown
to allow direct characterization of the initial nuclear motions which follow
electronic excitation.

1. Introduction

Molecular photodissociation is by now a venerable but still highly active arm of molecular
reaction dynamics. Most activity in this area is centred on the reactions of neutral molecules
using visible or near ultraviolet laser light sources; lasers are used for photodissociation itself,
for state-selective preparation of reactants and for characterization of the products. Recent
advances have allowed individual quantum state selection of reactants and determination of
the state distribution, orientation and alignment of products. Recent reviews [1, 2] give a
picture of a mature but still evolving field, with coherent control of reaction outcomes as a
tantalizing current goal [3, 4].
Despite the enormous power of laser techniques, most photodissociation work is still
devoted to two-body reactions, that is, to reactions in which a parent molecule breaks into just
two pieces. The dynamical characteristics of the products in such reactions are kinematically
controlled, so that by measuring the motions and internal state of a single fragment, a researcher
0953-4075/05/090839+21$30.00 © 2005 IOP Publishing Ltd Printed in the UK S839
S840 K Ueda and J H D Eland

can acquire complete knowledge of the dynamics of the photodissociation event. If a molecule
breaks into three or more fragments, the information from a single fragment only, however
precise, is insufficient to fully characterize the reaction. Only by measuring the motions of
at least two out of three, or in general n − 1 out of n products of a dissociation, detected in
coincidence, can one fully specify the reaction dynamics. It is in this domain of multi-body
dissociations (‘molecular explosions’) that vacuum ultraviolet (VUV) and x-ray studies of
photodissociation come into their own.
For photon energies from around 10 to about 30 eV, the major outcome of photoabsorption
is single ionization; from 40 eV onwards, double ionization and later triple and higher
ionization become dominant. The products of a photodissociation in this energy region
are positive ions, which are easily detected as single particles; using the fact that all products
are born at the same instant, all the ionic products of each single reactive event can be detected
and correlated. This coincidence technique is at the heart of modern photodissociation studies
in the VUV and x-ray realms, and gives access to kinematically complete descriptions of the
most basic of chemical reactions. The necessary light sources are provided in the laboratory
as fixed energy lines from atomic discharge lamps going up to about 50 eV photon energy
[5]. For tuneable light sources and for all higher energies, modern work relies on synchrotron
radiation from the many electron or positron storage rings available at national and international
facilities [6].
Measurement of the time of arrival of each charged particle at a detector is the most
fundamental requirement of the coincidence method; besides identifying correlated particles
it gives access to their velocity in one dimension. Although this is not sufficient for complete
reaction characterization, distributions of the correlated one-dimensional velocities of two or
more fragments from multi-particle events give a surprising amount of information on the
photodissociation reaction mechanisms. The technique known as photoelectron–photoion–
photoion coincidence (PEPIPICO) [7], introduced in 1986 and still in use in 2004, can be used
to distinguish correlated mechanisms from sequential or stepwise multi-body dissociations [8]
and to determine the various kinetic energy releases. It gives limited information on angular
distributions, however, and many deductions about mechanisms are left speculative. The
modern era of multiparticle coincidence studies of photodissociation, to which we devote the
remainder of this review, may be said to date from the addition of position-sensitive detectors
to the armoury of coincidence techniques, in 1995 [9].
As explained in more detail below, a position-sensitive detector (PSD) which also gives
accurate particle arrival times, allows all three momentum components of each particle’s
motion to be determined. The most widely used PSDs, which rely on television camera
technology, are unsuitable for this application because they are not sensitive to individual
arrival times and cannot distinguish particles which arrive in close temporal proximity. The
development and improvement of multi-hit PSDs based on delay-line technology has been a
crucial step in recent advances [9, 10]; other technologies now also promise similar capabilities
[11, 12].
In the first applications of PSD-PEPIPICO, molecules were ionized by temporally
continuous light sources, and electrons were detected to signal the instant of ionization, without
electron energy analysis. When this method is applied to dissociative double ionization of
diatomic molecules, the kinematical parameters are over-determined for the heavy particles
from a stationary point source and the extra information can be used to eliminate the Doppler
effect of target molecule motion. High resolution for kinetic energy release is obtained in
this way, and determines the spectrum of initial states, though uncertainty remains about the
internal states of the product atoms. Results for N2 , CO, NO and O2 showed that the double
photoionization processes at 40.8 and 48.4 eV photon energy are largely indirect for all but N2
Molecular photodissociation S841

[13]. (In an indirect double ionization, two electron ejections occur as separate stages; nuclear
motion between the stages is a possibility.)
The same technique, as applied in the laboratory to dissociative double ionization of the
tri-atomic and larger molecules [9] produced the first detailed trajectory studies of three-body
dissociations, including the generation of Newton diagrams. Later work on the tri-atomics SO2 ,
CS2 , OCS, ICN, CO2 and N2 O gave considerably more detail on the three-body dissociation
mechanisms and suggested that in the three-body dissociative double ionizations at 48.4 eV,
electron and nuclear motions indeed occur on competitive time scales [14]. The existence of
such processes has since been confirmed more directly by electron pair spectroscopy [15]. The
PSD-PEPIPICO technique was also applied to dissociative double ionization of the polyatomic
Fe(CO)5 molecule where detailed mechanistic pathways could be identified [16].
These early studies, sophisticated as they were, suffered from a lack of specificity about
the states of the multiply charged ions initially populated. This lack could be made up only
by energy analysis of the ejected electrons; one technique by which this can be partially
achieved is by provision of a second PSD for the electrons and use of the velocity imaging
method of Eppink and Parker [17]. The combined technique called VIPCO (velocity imaging
photoionization coincidence [18]) is particularly suited to study angular correlation between
ejected electrons and fragment ions. In cases where the heavy fragments recoil axially, that
is before any molecular rotation and with sufficient kinetic energy that the tangential velocity
from rotation can be ignored, the relative directions of electrons and fragment ions can be
transformed into fixed-molecule (electron) angular distributions. These angular distributions
are of great theoretical and topical interest, so most experiments using the VIPCO technique
have concentrated on measuring them rather than on the dissociation dynamics (see, for
example, Dowek et al [19], and references therein).
Interesting exceptions are seen in studies of photodissociation reactions where one of
the products is a negative ion, which can be detected in the same way as an electron with
determination of the initial velocity vector. The experiments give new information on simple
two-body (negative plus positive, ion) and three-body (negative, positive, neutral) reactions of
this type and in addition have discovered a new species of three-body photodissociation where
the products are two positive ions, one negative ion and one electron [18]. Such processes
may be rather common, but their partial cross-sections are always small by comparison with
simple electron ejections. In a related study, a photodissociation which appeared to be a simple
dissociative ionization was shown to involve the short-lived N− ion as an intermediate [20].
In the dissociative photoionization of CF4 and CCl4 , both orientation and alignment of the
products were found relative to the photoelectrons [21]. This surprising result was attributed,
like several phenomena already mentioned, to motion of the atomic nuclei on the same time
scale as ejection of the electron.
Interactions of nuclear and electronic motion are a recurrent theme in all the recent
experimental work on photodissociation in the VUV photon energy region, discussed so far.
Such interactions are clearly widespread and of great interest. To study them in detail it is
highly advantageous to move to the x-ray domain where the initial photon absorption produces
an inner shell vacancy. Benefits are the clear characterization of the initial state, where the
symmetry and lifetime are known, and the high degree of ionization finally produced; this
often allows detection and kinematic characterization of all products, even from complete
atomization. The first experiment of this kind was done for triple photoionization of SO2
at the Super-ACO storage ring light source at photon energies near the S 2p edge by the
PSD-PE3PICO (multiple-ion coincidence momentum imaging) technique [22]: the authors
inferred an interaction between nuclear and electronic motion similar to the valence double
photoionization.
S842 K Ueda and J H D Eland

21.5 71 140

80φ

Square- Drift Tube Hexagonal-


Type PSD Type PSD
Supersonic Jet

Figure 1. A schematic illustration of the multiple coincidence momentum imaging apparatus


(from [28]).

It can often happen that the core excited states of polyatomic molecules have an
equilibrium geometry different from the one of the neutral ground state. As a consequence,
nuclear motion may be brought about in the core excited states, during their femtosecond
lifetimes, before the system undergoes Auger decay followed by an ionic fragmentation.
The molecular deformation caused by the asymmetric nuclear motion is of particular interest
because it plays an important role in the decay processes of core excited states, and may
even open up new dissociation channels that are not reachable in the initial geometry
of the neutral molecule. The nuclear motion in the core excited state is reflected in
characteristics of the resonant Auger electron emission which takes place in competition with
the nuclear motion. Resonant Auger electron–ion coincidence is a suitable tool to investigate
such nuclear dynamics [23, 24]. Combining high-resolution resonant Auger spectroscopy
and ion momentum imaging in coincidence has been achieved very recently by some
groups [25–27].
In the rest of this paper, we first describe the two typical experimental techniques
introduced above, multiple-ion coincidence momentum imaging and resonant Auger electron–
ion momentum imaging, and then examine some individual results on photodissociation
following core-hole production, which illustrate how the complex multidimensional data
produced can be represented and interpreted.

2. Experimental setups

2.1. Multiple-ion coincidence momentum-imaging


The PSD-PEPIPICO, or multiple-ion coincidence momentum-imaging, is based on the time-
of-flight (TOF) method combined with a two-dimensional (2D) position-sensitive detector.
The apparatus installed in the beamline 27SU of SPring-8, the synchrotron radiation facility
in Japan, is schematically described in figure 1 [28] as a typical example. Consider the case
where one detects all the fragments in the form of atomic ions produced from the multiply-
charged parent molecular ions created via rapid multiple Auger decay. Registration of all
the fragments allows one to extract all the kinematical information for the linear momentum
(Px , Py , Pz ) of each fragment ion without ambiguity. A sample gas is introduced into the
source volume of the imaging apparatus, in the form of a supersonic jet, and crosses the
photon beam perpendicularly. The TOF axis is fixed in a horizontal direction and crosses both
the light beam and the molecular beam at right angles. The ion extraction field is typically
∼20 V mm−1. The length of the acceleration region is typically ∼70 mm, and the drift
Molecular photodissociation S843

8 channel Time to
Hexagonal delay line Amplifier & CFD digital converter
detector (HEX-80)
X1
X2

Y1
Y2 Personal
computer
Z1 T
D
Z2 C
DLATR6/Hex |
8

Ion MCP CFD

Amplifier
Electron DLATR6
MCP (amplifier & CFD)
Common
stop

Electronics for
common stop
generation

Figure 2. Schematic diagram of electronic system for recording the TOFs and the positions of
ions (from [31]).

region is about twice as long as the acceleration region, so as to satisfy the Wiley–Maclaren
space-focusing condition [29]. The TOF mass spectrometer is equipped with a multi-hit
position-sensitive detector. It consists of microchannel plates with effective diameter of 80 mm
and a hexagonal-type delay-line anode (Hex80, Roentdek [30]). This recently developed three-
layer delay-line anode allows us to register positions and TOFs of more than two ions with the
same mass in coincidence without suffering from the dead time intrinsic to the conventional
square delay-line anode.
The electronic system for recording the TOFs and positions of ions is schematically
illustrated in figure 2 [31]. There are eight output signals from the spectrometer; six from
the delay lines of the HEX-80 delay-line anode for ions, one from the MCP for ions and one
from the MCP for electrons. The signals from the delay lines are amplified by the differential
amplifier and shaped by constant fraction discriminators (CFDs) (RoentDek DLATR6/Hex).
The signals from the MCP for electrons are amplified by an amplifier and a CFD generates
a standard nim-pulse as an output. These eight shaped signals from the CFDs are put into a
time-to-digital converter (RoentDek TDC-8) as a start, which has a resolution of 500 ps, a full
scale of 32 µs, and 16-fold multi-hit capability. The TDC-8 is operated in a common stop
mode. When we detect one electron and two ions at least, a common stop signal is put into
the TDC-8. This common stop signal is produced from the electron and ion MCP signals with
the combination of fan-in-out, logic unit and gate-and-delay modules.
Note that the apparatus in figure 1 has a position-sensitive detector also on the opposite
side. Thus one can use it also for the VIPCO-type experiments [32–34].
S844 K Ueda and J H D Eland

ion tof spectrometer


SES2002 ion
delay-line
detector

+
apply pulsed electric field
electron after the detection of the
delay-line electron
detector
period 23 ns
bunch
marker HV
pulse
CFD
wait for
bunch and
marker start and
pulse CFD
generator ion
detection

or gate/delay

electron
random marker 4x and
marker

multi hit
common stop
start

TDC module

Figure 3. A schematic diagram of the experimental setup for high-resolution electron momentum-
resolved ion coincidence, including the data acquisition system. All signals from the two delay-line
detectors are fed into a 16-channel TDC system consisting of two TDC-8s (from [35]). See the
text for the details.

2.2. Electron momentum-resolved ion coincidence

High-resolution resonant Auger electron momentum-resolved ion coincidence can be achieved


by a combination of a high-resolution electron spectrometer and an ion momentum imaging
spectrometer [25–27]. Figure 3 is a schematic of the setup, including the data acquisition
system, installed in beamline 27SU at SPring-8 [35]. The electron and ion spectrometers are
mounted inside a vacuum chamber. The sample gas is introduced between the pusher and the
extractor electrode of the ion spectrometer through a grounded copper needle. After ionization
of the sample molecule, some of the electrons pass the pusher electrode and enter the electron
analyser (Gammadata Scienta SES-2002). In order to make coincidence experiments possible,
the standard CCD camera of the electron spectrometer was replaced by a delay-line detector
with effective diameter of 40 mm (Roentdek DLD40). During the coincidence experiment
all voltages of the electron spectrometer were fixed. Four channels of TDC-8 were used for
the recording of the position information of the electron detection. The common start for the
TDC-8 comes from the electron detector.
Triggered by the electron detection, rectangular high voltage pulses with opposite sign
are applied to the pusher and extractor electrodes. The drift tube is held at a constant voltage.
Molecular photodissociation S845

The electrodes of the ion spectrometer have 90% transmission copper grids (8 lines mm−1)
to produce homogenous electric fields between the different electrodes. The timing, quality
and stability of the high voltage pulses are crucial for the measurement. Especially the time
between the trigger signal and the output of the HV pulse must not exceed some 100 ns,
because otherwise the ionic fragment might leave the region between the pusher and extractor
electrode and thus be lost to the detection. The electronic insertion delay of the pulse generator
(GPTA HVC-1000 [36]) is less than 120 ns. The rise/fall (10%/90%) time is less than 15 ns.
The ions are detected by another delay-line detector with an active diameter of 80 mm
(Roentdek DLD80). Four channels of the second TDC-8 are used for the recording of the
position information of the ion detection. The ion side of the detector needs electronic gates
to mask the noise induced by the high voltage pulse generator.

3. Dissociation after selective core excitation of molecules

3.1. Nuclear motion in the Renner–Teller states of C 1s excited CO2


The structure of the CO2 molecule in the ground state belongs to D∞h and the electronic
configuration is
1σg2 1σu2 2σg2 3σg2 2σu2 4σg2 3σu2 1πu4 1πg4 ; 2πu0 5σg0 4σu0 .
Here 1σg and 1σu consist of symmetry-adapted linear combinations of the two O 1s orbitals
and 2σg is the C 1s orbital. There are three vibrational modes; the symmetric stretching mode
ν1 of σg symmetry, the bending mode ν2 of πu symmetry and the anti-symmetric stretching
mode ν3 of σu symmetry. A promotion of the C 1s core electron to the unoccupied molecular
orbital 2πu leads to the doubly degenerate u state that splits into two Renner–Teller (RT)
states along the bending coordinate Q2 [37–39]. The lower branch has a stable bent geometry
whereas the upper branch remains linear. We will demonstrate here that one can separate these
two bent and linear states by examining the direction of the bending nuclear motion brought
about in these states, by use of the multiple-ion-coincidence momentum imaging [31, 40, 41].
We consider here the triply charged molecular parent ion CO3+ 2 produced by the Auger
decay of the core excited states of CO2 . This triply charged molecular ion has high internal
energy and breaks up very rapidly by Coulomb explosion. The bond breaking is simultaneous
and the axial-recoil approximation [42–44] is valid. Thus linear momenta of the fragment ions
reflect the geometry of the molecule at the time when the Auger decay takes place. The triply
charged parent ion CO3+ 2 may be constructed via triple Auger decay and/or very fast sequential
decay (of the order of 10−15 s). Although the branching ratio for the production of the triply
charged molecular parent ion CO3+ 2 is only a few per cent, its dissociation characteristics
provide a good representation of the core excited molecule because the properties of the
excited state are independent of decay channels.
The bending motion in the A1 (B1 ) state proceeds in the direction parallel (perpendicular)
to the electric vector E of the light. Thus we can select the A1 state in which the direction of the
bending motion is parallel to E or the B1 state in which the molecular plane is perpendicular
to E , by examining the linear momenta of the three fragment ions C+ , O+ and O+ measured in
coincidence by means of the multiple-ion-coincidence momentum imaging.
The vector correlation among the linear momenta of the three fragment ions produced
in the three-body breakup CO3+ 2 → C + O + O can be well represented by the Newton
+ + +

diagrams as shown in figure 4 [41]. Because of momentum conservation, the sum of all
momenta is always zero and the three-particle fragmentation takes place in one plane. So we
can rotate all momenta in such a way that they are in the plane of the paper and that the first O+
S846 K Ueda and J H D Eland

1.0 (a) 1.0 (c)


+
C
0.5 0.5

0.0 0.0
+
O (1)

-0.5 -0.5
+
O (2)
-1.0 -1.0
p (C , O )
+
+

-1.0 -0.5 0.0 0.5 1.0 -1.0 -0.5 0.0 0.5 1.0
1.0 (b) + +
1.0 (d)
p(C ,O ) +
p(C ,O )
+

0.5 0.5

0.0 0.0

-0.5 -0.5

-1.0 -1.0

-1.0 -0.5 0.0 0.5 1.0 -1.0 -0.5 0.0 0.5 1.0
+ +
p (C , O )

Figure 4. Newton diagrams for the three-body breakup CO3+ 2 → C + O + O : (a) for the
+ + +

excitation at the C 1s → σ ∗ shape resonance (312 eV), (b) for the C 1s → 2πu excitation
(290.8 eV), (c) for the excitation to the state with the bending motion perpendicular to the electric
vector E (i.e. B1 in C2v ), by 290.8 eV photons, and (d) for the excitation to the state with the
bending motion parallel to E (i.e. A1 in C2v ), by 290.8 eV photons. The scale of contour plots is
linear and relative (from [41]).

goes exactly to the left and that the C+ has a positive y-momentum component. All momenta
are normalized to the momentum of the first O+ . The intensity is given in the form of contour
plots on a linear scale.
Compare diagrams (a) and (b) recorded at the C 1s → σ ∗ shape resonance at 312 eV
and at the C 1s → 2πu excitation at 290.8 eV, respectively. The C+ ion goes off slightly
transverse to the two O+ ions on average even at the C 1s → σ ∗ shape resonance, although the
molecule is believed to have a linear geometry before dissociation. This is because Coulomb
explosion enhances the small displacement in the zero point bending motion, resulting in
a non-negligible linear momentum of the C+ ion transverse to the two O+ ions. Note also
that the Newton diagram is constructed from the measurements for all solid angles. In the
procedure of projecting the vector correlation into the reaction plane, one implicitly introduced
the weighting factor of sin θ , where θ is the correlation angle between the two linear momenta
of the two O+ ions. Thus one can hardly see the completely linear correlation (θ = 180◦ )
between the two O+ ions owing to the weighting factor of sin θ . At the C 1s → 2πu excitation,
one can see the long tail for each island, which illustrates that the C+ ion goes off transverse
to the two O+ ions more than at the C 1s → σ ∗ shape resonance.
Molecular photodissociation S847

Compare diagrams (c) and (d). In both diagrams, the excitation energy is set at the
C 1s→ 2πu excitation peak. In the construction of diagram (c), however, the events
 are selected
 
in which the vector product of the two linear momenta of the two O+ ions, p O+(1) × p O+(2) ,
 +   + 
is parallel to E , with the angles between p O(1) × p O(2) and E smaller than 20◦ . In this
way, the B1 state in the bent geometry of C2v point group is selected. In constructing diagram
(d), on the other hand, the events are selected in which p(C+ ) is parallel to E , with the angles
between p(C+ ) and E smaller than 20◦ . In this way, the A1 state in the bent geometry of C2v
point group is selected.
Diagram (c) coincides with diagram (a) and illustrates that the B1 state thus probed has a
linear geometry. In diagram (d), one can clearly see that a very long tail appears for each island,
illustrating that the C+ ion goes off significantly transverse to the two O+ ions. This indicates
that the bending motion proceeds to a definite extent in the A1 state. The initial vibrational
wave packet has little momentum. The force to push the C atom at right angles relative to the
molecular axis must come, with time evolution, from the electronic potential energy surface.
Thus the state with A1 symmetry probed here must have a stable bent geometry.

3.2. H+2 formation mediated by the cis bending motion in the core excited HCCH molecule

The acetylene molecule HCCH is linear in the ground state and belongs to the D∞h point
group. The electronic configuration is

1σg2 1σu2 2σg2 2σu2 3σg2 πu4 ; 1πg0 3σu0 .

Here 1σg and 1σu consist of symmetry-adapted linear combinations of the two C 1s orbitals.
There are five vibrational modes: the C–H symmetric stretching mode ν1 of σg symmetry,
the C–C symmetric stretching mode ν2 of σg symmetry, the C–H antisymmetric stretching
mode ν3 of σu symmetry, the trans bending mode ν4 of πg symmetry and the cis bending
mode ν5 of πu symmetry. A promotion of the 1σu core electron to the unoccupied molecular
orbital 1πg leads to the doubly degenerate u state. This state splits into two Renner–Teller
(RT) states along both trans and cis bending coordinates Q4 and Q5 [45]. It is known that
the excited/ionized HCCH molecule can easily change its geometry to vinylidene, CCH2
[46]. We discuss the nuclear motion in the core excited HCCH molecule, which mediates H+2
formation [47].
To probe the nuclear motion in HCCH in the C 1s 1σu−1 πg excited state, we have employed
triple-ion-coincidence momentum imaging and detected H+ , H+ and C+2 from HCCH3+ parent
ions created after rapid multiple Auger decay. The vector correlation among the linear
momenta of the three fragment ions is shown in the Newton diagrams of figure 5. We rotate all
momenta in such a way that they are in the plane of the paper and that the first H+ goes exactly
to the left and that the other H+ has a positive y-momentum component. The excitation
energies are set at (a) above the C 1s ionization threshold (the monochromator is set to
270 eV but the second order light, 540 eV, is the dominant component) and (b) the C 1s
1σu−1 1πg excitation peak (285.8 eV). It is clearly seen in figure 5 that the intensity of the
second H+ around the position ‘A’, the negative x and positive y region, is higher in the
C 1s 1σu−1 1πg excited state than in the C 1s−1 ionized state. The events giving H+ intensity
in the region ‘A’ correspond to the processes in which the two H atoms approach each other
through the cis bending motion and then Coulomb explosion takes place. This suggests that
the cis bending motion is brought about in the C 1s 1σu−1 1πg core excited state. The trans
bending motion may also be brought about in the C 1s 1σu−1 1πg core excited state but the
deformation of the excited molecule to a bent configuration through the trans bending motion
S848 K Ueda and J H D Eland

(a) C 1s-1 H+
1
A
H+
0
Relative momentum
-1 C2 +

(b) C 1s-1π* H+
1
A
H+
0

C2 +
-1

-1 0 1 2
Relative momentum

Figure 5. Newton diagram for the three-body breakup HCCH3+ → H+ + H+ + C+2 (a) for the
C 1s−1 ionization measured at the photon energy of 270 (540) eV and (b) for the C 1s 1σu−1 1πg
excitation at the photon energy of 285.8 eV (from [47]).

would result in making the two H+ ions depart in directions opposite to one another. Thus the
Newton diagram is expected to be the same as that from a linear geometry.
H+2 molecular ions are produced only in the C 1s 1σu−1 1πg excited state. The ratios of ion
pair H+2 –C+2 to CH+ –CH+ are 1.7% at the C 1s 1σu−1 1πg excited state and 0.1% at the C 1s−1
ionized state. This observation suggests that H+2 molecular ions are not produced from CCH2 .
The cis bending motion in the C 1s 1σu−1 1πg excited state brings one of the H atoms in the
HCCH molecule close to the other H atom. The classical turning point of the cis bending
motion is at a bond angle of around 30◦ . This motion may reasonably be considered to unite
the two H atoms in HCCH. Thus only the cis bending motion mediates the formation of the
H+2 molecular ion, bringing the two terminal H atoms closer to one another.

3.3. Asymmetric nuclear motion in core excited/ionized BF3 molecules


The BF3 molecule has a D3h planar geometry in the ground state and its electronic configuration
is
(core) 1a2 4 2 4 2 4
1 1e 2a1 2e 1a2 3e 1e
4
1a2 0 0
2 ; 2a2 4e .
Here ‘core’ includes F 1s and B 1s electrons. The lowest unoccupied molecular orbital has 2a2
symmetry and B 2pz character, and is non-bonding. The 4e’ unoccupied molecular orbital has
antibonding character. There are four vibrational modes; the symmetric stretching (breathing)
mode ν1 of a1 symmetry, the out-of-plane bending mode ν2 of a2 symmetry, the asymmetric
stretching mode ν3 of e symmetry and the asymmetric wagging mode ν4 of e symmetry.

3.3.1. Out-of-plane motion in the B 1s excited BF3 . We consider the excitation of the B 1s
electron to the lowest unoccupied molecular orbital 2a2 . The B 1s−1 2a2 excited state has a
trigonometric pyramidal equilibrium geometry of C3v point group symmetry [48, 49]. Thus
after the core excitation, the central B atom is expected to go off transverse to the molecular
Molecular photodissociation S849

0.6

0.4

e1-1/3
0.2
e2 e3
0.0
e1
-0.2

-0.6 -0.4 -0.2 0.0 0.2 0.4 0.6


(e2-e3 )/31/2

Figure 6. Definition of the Dalitz plot (from [50]). See the text for details.

plane towards the new equilibrium geometry, via the out-of-plane bending vibration ν2 of
a2 symmetry. It is shown that a snapshot of the deformed core excited BF3 molecule can
be taken at the time of the Auger decay by use of the multiple-ion-coincidence momentum
imaging [50].
The quadruply charged BF4+ 3 molecular parent ion can be produced by the multiple Auger
decay of the core excited or ionized states of BF3 . This channel is very weak, representing less
than 1% of the total ionization yield. This quadruply charged molecular ion has high internal
energy and breaks up very rapidly by Coulomb explosion.
In the previous subsections we introduced the Newton diagram in order to represent the
vector correlation among the linear momenta of three ions produced by three-body breakup.
The alternative way to represent it is to use Dalitz plots [51]. Though Dalitz plots may
be less familiar than the Newton diagram, they have advantages in some cases because the
three particles are treated equivalently and each point of coordinates has equivalent weight in
the space of Dalitz plots. Figure 6 helps to explain the way to construct the Dalitz plot. We
introduce the normalized, squared momentum i for each ion i,
|Pi |2
i =  , (1)
i |Pi |
2

where Pi is the linear momentum of the ion i, and define the Cartesian coordinates xD and
yD as
 2 − 3 1
xD = 1/2
, yD = 1 − . (2)
3 3
Then all the data points (x, y) are within the circle in figure 6 and the distances from the data
point (x, y) to the three sides of the regular triangle are i .
In order to use the Dalitz plots for investigating the vector correlation among the four linear
momenta of the four particles produced by the four-body breakup, we take the projection of the
three linear momenta of the three particles to the plane perpendicular to the linear momentum
of the remaining fourth particle. First, consider the projections of the linear momenta of
the three F+ ions to the plane perpendicular to the linear momentum of the B+ ion, i.e. the
molecular plane, measured in coincidence. Figures 7(a) and (b) present the Dalitz plots for the
projected linear momenta of the three F+ ions thus defined; (a) is recorded at the B 1s → 4e
shape resonance above the B 1s ionization threshold, whereas (b) is recorded at the B 1s →
2a2 excitation. In both cases, most of the events are located near the central part of the Dalitz
S850 K Ueda and J H D Eland

0.3 (a) (c)


B 1s −> 4e'
0.2 B 1s −> 4e'

0.1

0.0

-0.1

-0.2
-1/3

-0.3

0.3
(b) (d)
1

B 1s −> 2a2''
0.2
B 1s −> 2a2''

0.1

0.0

-0.1

-0.2

-0.3

-0.3 -0.2 -0.1 0.0 0.1 0.2 0.3 -0.3 -0.2 -0.1 0.0 0.1 0.2 0.3

(ε2-ε3)/31/2

Figure 7. Dalitz plots for the three F+ fragments produced from BF4+ 3 : (a) recorded at the
B 1s → 4e shape resonance and (b) recorded at the B 1s → 2a2 excitation; projections are taken
for the linear momenta of the three F+ ions to the plane perpendicular to the linear momentum of
the B+ ion, recorded in coincidence. Dalitz plots for the B+ and two of the three F+ fragments

produced from BF4+ 3 , (c) recorded at the B 1s → 4e shape resonance and (d) recorded at the B
1s → 2a2 excitation; projections are taken for the linear momenta of B+ and two F+ onto the
plane perpendicular to the linear momentum of the remaining F+ ion, recorded in coincidence
(from [50]).

plots. This plot confirms that the projected linear momenta of the three F+ ions are almost the
same and thus the three-fold axial symmetry along the linear momentum of B+ is retained in
the four-body breakup, suggesting that the rupture of the three B+ –F+ bonds is simultaneous.
We now investigate the expulsion of the B+ fragment perpendicular to the molecular
plane, in order to obtain evidence for the out-of-plane ν2 nuclear motion. Let us consider the
projections of linear momenta of the B+ ion and the two F+ ions to the plane perpendicular to
the linear momentum of the remaining F+ ion. Figures 7(c) and (d) present the corresponding
Dalitz plots recorded at the B 1s → 4e shape resonance and at the B 1s → 2a2 excitation,
respectively. In the plot (c), most of the data points are close to the bottom. This clearly
indicates that the linear momentum of the B+ ion is small and the ionic dissociation is mainly
within the molecular plane. The non-zero component of the B+ linear momentum is small
but still finite. Even though the stable geometry of the B 1s−1 4e state and of the B 1s−1
ionized state is planar, the B atom goes off slightly transverse to the molecular plane due to
the zero-point ν2 vibration. Coulomb explosion enhances this small displacement, resulting in
a non-negligible component of the linear momentum of the B+ ion transverse to the molecular
plane.
Consider next the data in (d), recorded at the B 1s → 2a2 excitation. It can be seen that
the island at the bottom is elongated towards the positive vertical axis and the intensity at the
central part of the Dalitz plot is greatly enhanced, suggesting that the linear momentum of the
Molecular photodissociation S851

(a) (b)
0.2
II II

e1-1/3
0.0
I
-0.2
II
-0.2 0.0 0.2 -0.2 0.0 0.2
(e2-e3)/3 1/2 (e2-e3)/3 1/2

Figure 8. Dalitz plots: (a) and (b), events recorded on top of the B and F 1s → 4e shape
resonances, respectively (from [52]). See the text for further details.

B+ ion is increased. We recall that the rupture of the three B+ –F+ bonds is symmetric and
simultaneous as seen in figure 7(b). Thus the increase in the B+ linear momentum shown in
figure 7(d) is a direct proof that the out-of-plane motion proceeds to a certain extent in the B
1s−1 2a2 core excited state before the Auger decay occurs.

3.3.2. Asymmetric stretching motion in the F 1s ionized BF3 . We focus next on symmetry
lowering of the F 1s ionized state of the BF3 molecule. Here one could expect that asymmetric
nuclear motion of e symmetry should arise due to vibronic coupling, i.e. pseudo-Jahn–Teller
mixing among the nearly degenerate F 1s ionized states, and thus the molecular symmetry
should be lowered from D3h to C2v , reflecting dynamical core-hole localization. It is shown
that this asymmetric nuclear motion, which has never been detected by other techniques, can
also be probed by the multiple-ion coincidence momentum imaging [52].
We construct Dalitz plots with the projection of the linear momenta of the three F+ ions
into the plane perpendicular to the linear momentum of B+ . In figures 8(a) and (b), we compare
the Dalitz plots thus constructed for the B 1s ionization and F 1s ionization of BF3 . Figure 8(a)
is the same as figure 7(a) and presented here as a reference for comparison. In the plot of
figure 8(b) for the F 1s ionization, we find that a considerable fraction of the data points appear
along one of the three ternary axes closer to one side of the triangle. These events correspond
to the momentum sharing in which the projection for one of the three F+ ions to the plane
perpendicular to the B+ ejection is almost zero. One can imagine that these events reflect
asymmetric stretching nuclear motion in the F 1s ionized state. We will demonstrate that this
is really the case. The masked areas I and II in figure 8(b) will be used later to gate symmetric
and asymmetric events, respectively.
In order to illustrate how the four ions fly apart, we employ here Newton diagrams. To
display correlations
 of four linear momenta for the four-body
  breakup,
  we define the plane
by P(B+ ) and P F+b and take the projections of P F+a and P F+c onto this plane. Then
the amplitudes of the four momenta within this plane are normalized in such a way that the
amplitude of P(B+ ) is unity. Figures 9(a) and (b) correspond to the Newton diagrams thus
produced for the symmetric and asymmetric events falling in the masked areas I and II in
figure 8(b), respectively.
 The data in figure 9(b) are further selected by requiring the angle
between P F+a and P(B+ ) to be larger than 150◦ . Here the unit vector corresponding to the
normalized P(B+ ) is set to the positive y direction, (x, y) = (0, 1).
From figure 9(a), it is clear that the symmetric events I are the breakup within the three-fold
symmetry, where B+ is ejected perpendicularly to the molecular plane and three F+ are ejected
nearly within the molecular plane. This breakup pattern reflects that the four-body breakup
of BF4+3 by Coulomb explosion starts at the conformation close to the ground state of D3h
S852 K Ueda and J H D Eland

2 (a) +
(b) +
P(B ) P(B )
1 P(Fc )
+

0 +
+
+ P(Fc ) P(Fb )
-1 P(Fb )
+ +
-2 P(Fa ) P(Fa )
-2 -1 0 1 2 -2 -1 0 1 2

Figure 9. Newton diagrams for the symmetric (a) and asymmetric (b) events, respectively. Here
the F+ that goes in the direction closest to the opposite to P(B+ ) is labelled as F+a and the other
two as F+b and F+c . The diagrams are constructed for the momenta projected to the plane defined
by P(B+ ) and P(F+b ) (from [52]). See the text for further details.
(This figure is in colour only in the electronic version)

symmetry. Figure 9(b), on the other hand, clearly illustrates that the asymmetric events (II)
are the breakup within the molecular plane, where B+ and F+a fly apart in opposite directions
and F+b and F+c fly apart in opposite directions perpendicular to B+ and F+a . This can happen
only when the four-body breakup by Coulomb explosion starts at the conformation in which
the central boron atom is off-centre in the direction of one B–F bond. Thus the enhancement
of the asymmetric events (II) by the F 1s ionization is direct proof that the asymmetric nuclear
motion, which brings the central B atom off-centre within the molecular plane, is initiated by
the F 1s ionization. Note that the multiple Auger decay may take place before the central B
atom moves sufficiently, even though the asymmetric nuclear motion is initiated in the F 1s
ionized state. Then four-body breakup by Coulomb explosion ends up with the symmetric
events I. Note also that the asymmetric events take place for the B 1s ionization as well (see
figure 8(a)) due to the zero-point asymmetric vibration, though the amount is far less than that
for the F 1s ionization.
Finally we note the correlation between the symmetry-lowering axis and the polarization
axis. The asymmetry parameters β of B+ and F+ for the symmetric four-body breakup events
I are −0.4 ± 0.1 and 0.4 ± 0.1, respectively. These β values suggest that the F+ (B+ ) is
ejected preferentially within (perpendicularly to) the molecular plane whose plane normal is
preferentially oriented perpendicularly to the polarization vector, due to the e character of the
shape resonance. The β parameters of B+ and F+a for the asymmetric events (II) are 0.7 ± 0.1.
These strongly positive values illustrate that B+ and F+a are emitted preferentially along the
polarization vector and thus there is direct evidence that the elongation of the B–F bond due
to asymmetric nuclear motion resulting in the D3h → C2v symmetry lowering takes place
preferentially along the electric vector E of the light.

3.4. Asymmetric ultrafast dissociation of F 1s excited CF4

The ground-state electronic configuration of a CF4 molecule which belongs to a Td point


group in its ground state is

(core) 1a21 1t26 2a21 2t26 1e4 3t26 1t16 ; 3a∗0 ∗0


1 4t2 , (3)

where ‘(core)’ includes F 1s and C 1s, 2a1 and 2t2 have C–F bonding character and the 3a∗1
and 4t∗2 unoccupied orbitals are counterparts of these two and have antibonding character.
Molecular photodissociation S853

(a)
3 B

Cross section (arb. units)


2

1
A
0
(b)

0
Anisotropy parameter β

0.4 (c)

0.2

0.0

686 688 690 692 694 696 698 700


Photon energy (eV)

Figure 10. (a) Total ion yield spectrum in the F 1s excitation region of the CF4 molecule.
(b) Yield spectra of energetic ions measured in the directions parallel and perpendicular to the
electric vector E of the light (solid and dashed lines, respectively). (c) Photoion anisotropy
parameter β extracted from the angle-resolved spectra. The photon energies used in the Auger
decay study are indicated by arrows on the total ion yield spectrum (from [53]).

3.4.1. Asymmetric ion ejection. In figure 10, we present ion yield spectra of the CF4
molecule near the F 1s ionization threshold [53]. The top panel shows the total ion yield
spectrum recorded by the ion detector for 4π sr collection, whereas the middle panel shows
the angle-resolved yield spectra, I (0◦ ) and I (90◦ ), of energetic fragment ions with kinetic
energy larger than 6 eV, recorded by the two identical retarding-potential ion detectors placed
in the direction parallel and perpendicular to the electric vector E of the light. Values of the
photoion anisotropy parameter β obtained from I (0◦ ) and I (90◦ ) are also plotted in figure 10
(bottom panel). The largest anisotropy of β ∼ 0.36 is around the weak shoulder structure A
at ∼690 eV.
We discuss the origin of the preferential dissociation along the electric vector E of the
light. To do so, we start with the symmetry-adapted molecular orbital picture [54, 55]. Then
the F 1s shell forms symmetry-adapted t2 and a1 molecular orbitals in CF4 . Dipole-allowed
transitions from F 1s a1 lead to t2∗ , whereas those from F 1s t2 lead to both a∗1 and t2∗ . In
analogy with the case of SiF4 [56], the main peak B in figure 10 may correspond primarily to
the F 1s a1 , t2 → t2∗ excitation, whereas the weak feature A is assigned to the F 1s t2 → a∗1
excitation. Introducing the molecular frame axis z that coincides with one of the four F–C
S854 K Ueda and J H D Eland

axes and x-axis in such a way that one of the other three F atoms lies in the x–z plane, we
can label the F atom on the z-axis F(1) , one on the x–z plane F(2) and the other two F(3) and
F(4) . Using the localized-hole wavefunctions i ≡ F(i) 1s−1 a∗1 as a basis set, we can express
symmetry-adapted wavefunctions of the core excited states F 1s a−1 ∗ ∗ −1 ∗ ∗
1 a1 A1 and F 1s t2 a1 T2
in the following manner:

A∗1 = 12 (1 + 2 + 3 + 4 ), (4)


T∗2 (x) = √1 (22
6
− 3 − 4 ), (5)

T∗2 (y) = √1 (3


2
− 4 ), (6)

T∗2 (z) = √1 (31


12
− 2 − 3 − 4 ). (7)

Photoabsorption from the A1 ground state to the three degenerate T∗2 states is allowed, while
the transition to A∗1 is dipole forbidden.
We can take the axis of the F+ ion ejection to be the F(1) –C axis (i.e. z-axis) without loss
of generalization. The ion detectors are mounted parallel (0◦ ) and perpendicular (90◦ ) to the
electric vector E of the light. F+ detection at 0◦ corresponds to events in which the F(1) –C axis
(z-axis) coincides with the electric vector E of the light. In this case, referring to equation (7),
we find that there is a significant possibility that the F(1) has a F 1s hole. On the other hand, the
F+ detection at 90◦ corresponds to the events in which the F(1) –C axis (z-axis) is perpendicular
to the electric vector E . Then, referring to equations (5) and (6), we find that there is no
possibility that F(1) has a F 1s hole. Thus the positive value of β 0.36 at the feature A in
figure 10 is attributed to the preferential rupture of the F–C bond along the axis where the
F core-hole is situated. It should be noted that even if asymmetric nuclear motion does not
occur in the core excited state, Auger decay can transfer the alignment to the Auger-final state,
where the asymmetric nuclear motion in the Auger-final state may result in anisotropic ionic
fragmentation. The ion yield measurement cannot distinguish symmetry breaking in the core
excited state from that in the Auger-final state.

3.4.2. Doppler effects in Auger emission from fragments. Figure 11 contains a series of
electron-emission spectra from the decay of core excited states in CF4 with energies close to
690 eV. The measurements were made at eight different photon energies indicated by arrows
in figure 10, at angles both parallel and perpendicular to the electric vector E of the light
[53]. The spectral feature P disperses linearly in kinetic energy as the photon energy is tuned
through the state; this feature corresponds to the D 2 A1 photoelectron band. The spectral
feature F at ∼656.2 eV, on the other hand, remains at the same kinetic energy in all spectra.
The broader features S1 and S2 also appear at fixed kinetic energies of about 654 eV and
658 eV. This non-dispersive behaviour stems from transitions between potential energy
surfaces which are locally parallel. At the limit of large distances the transition is between
levels in a fragment. The excess kinetic energy is transferred to kinetic energy of the fragment.
Looking more closely at peak F, we find that the spectra measured around 690 eV show a
double peak structure at 0◦ . This behaviour is indicative of the Doppler-like splitting measured
previously in, for example, O2 [57]. If the dissociation of the molecule occurs on a time scale
comparable to the Auger decay, and takes place preferentially along the electric vector E of
the light, then electron emission can be observed with Doppler shifts due to the motion of
the fragment that emits the electron. The Doppler shift of the electron kinetic energy for the
fragments propagating in opposite directions has the opposite sign. The double peak structure
in the 0◦ spectrum corresponds to this Doppler splitting. When dissociation takes place
Molecular photodissociation S855

S1 F
P
S2

hν = 691.2 eV

690.8

690.4
Intensity

690.0

689.6
X2

689.1
X 2.5

688.7
X5

688.3 X 10

655 660 665


Kinetic energy (eV)

Figure 11. Angle-resolved electron spectra recorded across the F 1s → a∗1 resonance for eight
different photon energies indicated by arrows in figure 10. The dotted line is the measurement
with the electron spectrometer lens mounted perpendicular to the electric vector E of the light, and
the solid line is the in-plane measurement (from [53]).

preferentially along the axis perpendicular to the observer, the direction of the dissociation is
symmetric with respect to the measurement direction. Thus the Auger electron emitted from
the departing fragment results in a single peak structure as seen in the 90◦ spectrum. The
observation of the Auger emission of the atomic-like F∗ fragment and its Doppler splitting in
the 0◦ spectrum is direct evidence that the asymmetric nuclear motion proceeds in the F 1s
excited state.
We now consider how this asymmetric nuclear motion is caused. The F 1s−1 a∗1 core
excited states consist of nearly degenerate A∗1 and T∗2 states. A coupling between the electron
motion and the asymmetric nuclear motion causes pseudo-Jahn–Teller mixing between these
A∗1 and T∗2 core excited states, in addition to the Jahn-Teller mixing within T∗2 . When one C–F
bond is elongated, the molecule belongs to the C3v point group. The T∗2 state splits into A∗1 and
E∗ components. The A∗1 couples with the other A∗1 which originates from A∗1 in the symmetric
CF4 geometry. As a result of these couplings, one of the A∗1 potential energy curves is expected
to be strongly repulsive along the asymmetric stretching coordinate. This mechanism leads
to fast fragmentation along one C–F axis and results in Auger electron emission from the
departing fragment.
One may also address the question of whether the observed atomic-like electron emission
necessarily takes place at the dissociation limit. The dissociation is not complete within the
F 1s hole lifetime in a strict sense. We will discuss this specific aspect below.
S856 K Ueda and J H D Eland

late ion
blue
red late ion

late ion
red
F ion
CF3 fragment

Figure 12. Schematic diagram of the processes detected in the present coincidence experiment.
Only the electrons ejected in the left direction are detected. Processes (i) and (ii) without scattering
can be detected as coincidence events between blue-shifted electrons and late ions, and between
red-shifted electrons and early ions, respectively, while process (iii) involving the back-scattering
can be detected as coincidence events between red-shifted electrons and late ions (from [58]).

3.4.3. Back-scattering of the Auger electron. A more detailed insight into the process is
possible by measurement of the emitted Auger electron in coincidence with the emitting
fragment [25, 26]. This type of coincidence experiment in fact provides evidence that a
significant fraction of the Auger electrons emitted from the atomic F fragment undergoes
back-scattering on the other fragment CF3 [58].
We consider only events where an electron and an F+ ion are detected in coincidence.
Different contributions to the events detected in the coincidence experiment are illustrated in
figure 12. In order for the electron to be detected, it must be emitted in the direction of the
entrance slit of the electron spectrometer, while the F∗ atom may fly in any direction. For
simplicity, only the F∗ atoms flying to the left and right are shown in the figure. These are
the most common cases, because the fragmentation preferentially takes place along the linear
polarization direction of the light which coincides with the direction of the electron detection
[53]. The processes (i) and (ii) involve only direct emission of the electron without scattering.
Thus the ‘red’ and ‘blue’ shifted electrons always correspond to a shorter or larger TOF of the
ions, respectively, where the TOF is compared with the TOF of a non-energetic F+ ion. The
ions are termed ‘early’ and ‘late’ accordingly.
Figure 13 shows an example of the data sets. Each point in the coincidence map (c)
corresponds to one electron–ion coincidence event. Below the F+ Auger peak in the electron
spectrum (a), two clusters of points are visible in the map (c). They belong to the processes
(i) and (ii) in figure 12, i.e., coincidence events between blue-shifted electrons and late ions
and those between red-shifted electrons and early ions, respectively. The diagonal line that
crosses these two clusters of points shows the expected correlation between the ion TOF and
the electron kinetic energy, which is calculated based on knowledge of the fields in the ion
spectrometer which gives the relation of the TOF and the ion momentum. Knowing the ion
momentum, the Doppler shift of the electron can be calculated. The fact that the events scatter
symmetrically around this line indicates that the electron is emitted when the ion has already
reached its final speed. Thus the Auger decay takes place in the atomic regime. An earlier
emission must lead to a smaller Doppler shift of the electron. Thus our observation is a direct
proof that the first phase of the fragmentation, the acceleration of the core-excited F∗ atom, is
faster than the Auger decay.
If a significant fraction of the electrons that are emitted in the direction of CF3 undergo
back-scattering, this will lead to a reduction of the count rate for process (ii) because the
Molecular photodissociation S857

60 electrons F 1s Auger

Counts
(a)
40
20

electron - F+ coincidences (c) (b)


4.2 late
F+ TOF (µs)
4.1 F+

early
4.0
30 (d) electrons
Counts

with early F+ with late F+ 50 100


20 Counts
10

650 652 654 656 658


Electron kinetic energy (eV)

Figure 13. (a) Electron spectrum coincident with F+ ions. (b) F+ ion TOF spectrum coincident
with electrons in (a). The black data points in (a) and (b) are all coincident counts. The grey lines
below are the random contributions. (c) Coincidence map. Each event is represented by a dot.
(d) Electron spectra that are coincident with the ‘early’ and the ‘late’ F+ ions, as illustrated in (b).
In this graph the contribution of the random coincidences has been subtracted (from [58]).

0.05

0.00
PB

0.05

0.10

689 690 691


Photon energy (eV)

Figure 14. Early/late asymmetry of the ion TOF peak PB as a measure of the strength of the
back-scattering of the Auger electron at the CF3 remainder. The different marks are from different
beam times using the same apparatus. The grey line is the weighted average of all data points
(from [58]).

electron is no longer detected and, instead, process (iii) in figure 12 is detected for electrons
that are initially emitted away from the electron detector and then reflected in its direction.
Process (iii) can be distinguished from (i) and (ii) as a coincidence signal between the red-
shifted electron and the late ion. The degree of the early–late asymmetry (the probability of
the back-scattering) may be defined as
Ie − Il
PB = , (8)
Ie + Il
S858 K Ueda and J H D Eland

where Ie and Il are the integrated intensity (area) of the early and late ions, as illustrated in
figure 13(b). The results for the early–late asymmetry are shown in figure 14. The back-
scattering is found to be significant: it has an average value of PB = −4.7% of the total
coincident intensity.

4. Conclusion

The work discussed in this review demonstrates how recent technical advances, in light
sources, position-sensitive detectors and their experimental deployment, have made molecular
photodissociation in the VUV and x-ray regions an extremely powerful probe of molecular
reaction dynamics. The initial nuclear motion following excitation, which is the first and
vital stage of dissociation, can now be studied directly. From the early work which identified
temporal competition between nuclear and electronic motion non-specifically, the newer core
excitation results have progressed to identify individual normal modes excited. The identities
of the normal modes relate clearly to the changes in potential energy surfaces brought about by
electronic excitation. The new experiments provide extremely rich data, with details not only
on the directions and symmetry character of the initial nuclear motions, but also on their time
scales. This power of the technique can be used to develop and test both chemical insights
and quantum-mechanical models of the photodissociation process.

Acknowledgments

The experimental results described as examples were taken at SPring-8 with approval of the
programme review committee via cooperative research projects. KU is grateful to Norio Saito,
Alberto De Fanis, Georg Prümper and a number of other collaborators who participated in a
series of cooperative projects for invaluable contributions and helpful discussion and to Japan
Society of the Promotion of Science for Grants-in-Aid for Scientific Research.

References

[1] Lee Y-P 2003 Annu. Rev. Phys. Chem. 54 215


[2] Yang X and Liu K (ed) 2004 Modern Trends in Chemical Reaction Dynamics (New Jersey: World Scientific)
[3] Gordon R J, Zhu L and Seidman T 1999 Acc. Chem. Res. 32 1007
[4] Gordon R J, Zhu L and Seidman T 2001 J. Phys. C: Solid State Phys. 105 4387
[5] Baltzer P and Karlsson L 1988 Phys. Rev. A 38 2322
[6] Winick H 1994 Synchrotron Radiation Sources: A Primer (Series on Synchrotron Radiation Techniques and
Applications vol 1) (London: World Scientific)
[7] Eland J H D 1987 Mol. Phys. 61 725
[8] Maul C and Gericke K-H 1997 Int. Rev. Phys. Chem. 16 1
[9] Hsieh S and Eland J H D 1995 Rapid Commun. Mass Spectrom. 9 1261
[10] Jagutzki O, Mergel V, Ullmann-Pfleger K, Spielberger L, Meyer U, Dörner R and Schmidt-Bocking H 1998
Proc. SPIE 3438 322
[11] Lavollée M 1999 Rev. Sci. Instrum. 70 2968
[12] Rajgara F A, Mathur D, Nishide T, Kitamura T, Shiromaru H, Achiba Y and Kobayashi N 2002 Int. J. Mass
Spectrom. 215 15
[13] Hsieh S and Eland J H D 1996 J. Phys. B: At. Mol. Opt. Phys. 29 5795
[14] Hsieh S and Eland J H D 1997 J. Phys. B: At. Mol. Opt. Phys. 30 4515
[15] Eland J H D 2003 Chem. Phys. 294 171
[16] Hsieh S and Eland J H D 1997 Int. J. Mass Spectrom. Ion Process. 167 415
[17] Eppink A T J B and Parker D H 1997 Rev. Sci. Instrum. 68 3477
[18] Hikosaka Y and Eland J H D 2000 Rapid Commun. Mass Spectrom. 14 2305
[19] Dowek D, Lebech M, Houver J C and Lucchese R R 2004 J. Electron Spectrosc. Relat. Phenom. 141 211
Molecular photodissociation S859

[20] Hikosaka Y and Eland J H D 2001 Chem. Phys. 272 91


[21] Kinugawa T, Hikosaka Y, Hodgekins A M and Eland J H D 2002 J. Mass Spectrom. 37 854
[22] Lavolee M and Brems V 1999 J. Chem. Phys. 110 918
[23] Ueda K, Simon M, Miron C, Leclercq N, Guillemin R, Morin P and Tanaka S 1999 Phys. Rev. Lett. 83 3800
[24] Morin P, Simon M, Miron C, Leclercq N, Kukk E, Bozek J D and Berrah N 2000 Phys. Rev. A 61 050701
[25] Kugeler O, Prümper G, Hentges R, Viefhaus J, Rolles D, Becker U, Marburger S and Hergenhahn U 2004
Phys. Rev. Lett. 93 033002
[26] Prümper G, Tamenori Y, De Fanis A, Hergenhahn U, Kitajima M, Hoshino M, Tanaka H and Ueda K 2005
J. Phys. B: At. Mol. Opt. Phys. 38 1
[27] Céolin D, Miron C, Simon M and Morin P 2004 J. Electron Spectrosc. Relat. Phenom. 141 171
[28] Saito N, Ueda K and Koyano I 2003 X-ray and Inner-Shell Processes: 19th Int. Conf. on X-Ray and Inner-Shell
Processes ed A Bianconi, A Marcelli and N L Saini AIP Proc. 652 172
[29] Wiley W C and McLaren I H 1955 Rev. Sci. Instrum. 26 1150
[30] See http://roentdek.com for details on the detectors and related electronics
[31] Saito N et al 2004 J. Electron Spectrosc. Relat. Phenom. 141 183
[32] De Fanis A et al 2002 Phys. Rev. Lett. 89 023006
[33] Saito N et al 2003 J. Phys. B: At. Mol. Opt. Phys. 36 L25
[34] Jahnke T, Weber Th, Osipov T, Landers A L, Jagutzki O, Schmidt L Ph, H, Cocke C L, Prior M H,
Schmidt-Böcking H and Dörner R 2004 J. Electron Spectrosc. Relat. Phenom. 141 229, and references
cited therein
[35] Prümper G, Ueda K, Hergenhahn U, De Fanis A, Tamenori Y, Kitajima M, Hoshino M and Tanaka H 2005
J. Electron Spectrosc. Relat. Phenom. at press
[36] See http://www.gpta.de for details on the HV pulse generator
[37] Wight G R and Brion C E 1973 J. Electron Spectrosc. Relat. Phenom. 3 191
[38] Adachi J, Kosugi N, Shigemasa E and Yagishita A 1997 J. Chem. Phys. 107 4919
[39] Yoshida H, Nobusada K, Okada K, Tanimoto S, Saito N, De Fanis A and Ueda K 2002 Phys. Rev. Lett. 88
083001
[40] Muramatsu Y et al 2002 Phys. Rev. Lett. 88 133002
[41] Muramatsu Y et al 2002 Surf. Rev. Lett. 9 93
[42] Zare R N 1972 Mol. Photochem. 4 1
[43] Busch G E and Wilson K R 1972 J. Chem. Phys. 56 3638
[44] Dehmer J L and Dill D 1978 Phys. Rev. A 18 164
[45] Kempgens B, Itchkawitz B S, Feldhaus J, Bradshaw A M, Koppel H, Doscher M, Gadea F X and
Cederbaum L S 1997 Chem. Phys. Lett. 277 436
[46] Osipov T, Cocke CL, Prior M H, Landers A, Weber Th, Jagutski O, Schmidt L, Schmidt-Böcking H and
Dörner R 2003 Phys. Rev. Lett. 90 233002
[47] Saito N, Nagoshi M, Machida M, Koyano I, De Fanis A and Ueda K 2004 Chem. Phys. Lett. 393 295
[48] Tanaka S, Kayanuma Y and Ueda K 1998 Phys. Rev. A 57 3437
[49] Simon M, Morin P, Lablanquie P, Lavollée M, Ueda K and Kosugi M 1995 Chem. Phys. Lett. 238 42
[50] Ueda K et al 2002 Chem. Phys. 289 135
[51] Dalitz R H 1953 Phil. Mag. 44 1068
[52] De Fanis A, Saito N, Machida M, Okada K, Chiba H, Cassimi A, Dörner R, Koyano I and Ueda K 2004 Phys.
Rev. A 69 022506
[53] Ueda K et al 2003 Phys. Rev. Lett. 90 233006
[54] Domcke W and Cederbaum L S 1977 Chem. Phys. 25 189
[55] Ueda K, Shimizu Y, Chiba H, Okunishi M, Ohmori K, West J B, Sato Y, Hayaishi T, Nakamatsu H and
Mukoyama T 1997 Phys. Rev. Lett. 79 3371
[56] Okada K, Tamenori Y, Koyano I and Ueda K 2002 Surf. Rev. Lett. 9 89
[57] Björneholm O et al 2000 Phys. Rev. Lett. 84 2826
[58] Prümper G, Ueda K, Tamenori Y, Kitajima M, Kuze N, Tanaka H, Makochekanwa C, Hoshino M and Oura M
Phys. Rev. A at press
[59] Kitajima M et al 2003 Phys. Rev. Lett. 91 213003

You might also like