Probing transient molecular structures with time-resolved pump/probe XAFS using synchrotron X-ray sources

Article abstract of DOI:10.1016/S0368-2048(01)00289-4

Laser pulse pump, X-ray pulse probe X-ray absorption fine structure (pump-probe XAFS) experiments using synchrotron sources are described from technical considerations and from scientific significance. There are three technical challenges of such experiments: (1) laser photoexcitation, (2) synchronization of laser pulse and X-ray pulse, and (3) detection; each of which is investigated in detail. Based on the results from these investigations, the transient molecular structure of a reaction intermediate produced by photoexcitation of NiTPP-L₂ (NiTPP, nickeltetraphenylporphyrin; L, piperidine) in solution has been captured for the first time with the pump-probe XAFS on a 14-ns time scale obtained from the X-ray pulses from a third generation synchrotron source. The experimental results confirm that photoexcitation leads to the rapid removal of both axial ligands to produce a transient square-planar intermediate, NiTTP, with a lifetime of 28 ns. The transient structure of the photodissociated intermediate is nearly identical to that of the ground state NiTPP, suggesting that the intermediate adopts the same structure as the ground state in a non-coordinating solvent before it recombines with two ligands to form the more stable octahedrally coordinated NiTPP-L₂. No detectable population of a penta-coordinated intermediate was present. This experiment demonstrates the feasibility of determining transient molecular structures in disordered media using the temporal resolution of a synchrotron X-ray source.

Full text of DOI:10.1016/S0368-2048(01)00289-4

Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 161–174
www.elsevier.nl/locate/elspec
Probing transient molecular structures with time-resolved pump/
probe XAFS using synchrotron X-ray sources
*
Lin X. Chen
Chemistry Division, Argonne National Laboratory, 9700 South Cass Avenue, Argonne, IL 60439, USA
Abstract
Laser pulse pump, X-ray pulse probe X-ray absorption fine structure (pump-probe XAFS) experiments using synchrotron
sources are described from technical considerations and from scientific significance. There are three technical challenges of
such experiments: (1) laser photoexcitation, (2) synchronization of laser pulse and X-ray pulse, and (3) detection; each of
which is investigated in detail. Based on the results from these investigations, the transient molecular structure of a reaction
intermediate produced by photoexcitation of NiTPP-L2 (NiTPP, nickeltetraphenylporphyrin; L, piperidine) in solution has
been captured for the first time with the pump-probe XAFS on a 14-ns time scale obtained from the X-ray pulses from a
third generation synchrotron source. The experimental results confirm that photoexcitation leads to the rapid removal of both
axial ligands to produce a transient square-planar intermediate, NiTTP, with a lifetime of 28 ns. The transient structure of the
photodissociated intermediate is nearly identical to that of the ground state NiTPP, suggesting that the intermediate adopts
the same structure as the ground state in a non-coordinating solvent before it recombines with two ligands to form the more
stable octahedrally coordinated NiTPP-L2. No detectable population of a penta-coordinated intermediate was present. This
experiment demonstrates the feasibility of determining transient molecular structures in disordered media using the temporal
resolution of a synchrotron X-ray source.
2001 Elsevier Science B.V. All rights reserved.
Keywords: XAFS; Pump-probe; Time-resolved XAFS; NiTPP; Synchrotron; Structures of intermediates
1. Introduction
ing to the Franck–Condon principle, the initial
reaction coordinates in the excited states remain the
same as those in the ground state despite changes in
the electron density distribution due to photoexcita-
tion [1]. The nuclear coordinates will respond to the
electron density change through various rapid relaxa-
tion processes, resulting in equilibrated excited state
reaction coordinates that differ from those of the
ground state (Fig. 1). In solution, the time required
for the initial excited state nuclear coordinates to
relax to the equilibrated excited state coordinates
could be as long as a few vibrational cycles, or on
femto to pico second time scales [2]. These excited
states can proceed to form other transient states and
the product, or they can return to the ground state.
Molecular structures are the bases of understand-
ing photochemical reaction mechanisms and molecu-
lar reactivity. Although many aspects of photo-
chemistry can be well understood based on the
ground state structures, our understanding of the
processes often remains incomplete due to unknown
structures of transient molecular species, such as the
excited states. When a molecule is excited by light,
its electron density distribution is changed. Accord-
*
Tel.: 11-630-252-3533; fax: 11-630-252-9289.
E-mail address: lchen@anl.gov (L.X. Chen).
0368-2048/01/$ – see front matter
PII: S0368-2048(01)00289-4
2001 Elsevier Science B.V. All rights reserved.

162
L.X. Chen / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 161–174
state, as well as for capturing the molecular struc-
tures of other transient species. Some experiments
have been carried out using bench-top laser plasma-
based and synchrotron X-ray sources [20–22]. Be-
cause many photochemical reactions occur in dis-
ordered media, X-ray techniques that do not rely on
the long-range order of the structure must be de-
veloped. Although the concept of laser pump/X-ray
probe X-ray absorption fine structure (XAFS) using
synchrotron radiation was proposed earlier [23–25],
transient molecular structures generated in light-in-
duced processes were mainly captured in the past
without using the temporal resolution of the source
[26–34].
We report here our effort in determining transient
molecular structures during photochemical reactions
by pump-probe XAFS using a third generation
synchrotron X-ray source. Because the intrinsic time
resolution of the pump-probe technique is deter-
mined by the pulse duration of the pump or the
probe, whichever is longer, the synchrotron X-ray
pulse will provide a time resolution of the order of
100 ps [35]. This time scale is apparently too long
for monitoring the atomic movement from the initial
Franck–Condon state to the equilibrated excited
state, but it is suitable for capturing structures of the
equilibrated excited states within their lifetimes
ranging from 100 ps to longer. The structural in-
formation on these equilibrated excited states, as
well as other transient species, is crucial to the
fundamental understanding of the photochemistry.
The third generation synchrotron X-ray sources
provide not only an X-ray brilliance several orders of
magnitude higher than previous X-ray sources, but
also novel timing structures of X-ray pulses [35].
The high X-ray brilliance and flux shorten the time
required to collect structural data and the timing
structures of the X-ray pulses allow extraction of the
probe signals from a single X-ray pulse. Thus, the fs
laser plasma-based and synchrotron-based X-ray
sources are complementary in solving transient mo-
lecular structures on different time scales. In the
following, we will discuss the challenges in the
pump-probe XAFS using the synchrotron X-ray
sources, and our recent work on the determination of
a transient molecular structure during a photodis-
sociation process.
Fig. 1. Molecule interaction with light in a reversible reaction.
The ground state (GS) molecules are excited to an excited state
(ES) which undergoes fast relaxation to the equilibrated reaction
coordinates qES, different from those of the GS, qGS. The ES may
proceed to other transient states (TS) or return to the GS directly.
Very little information on qES and qTS is known. The pump-probe
XAFS aims to determine these structures during their short
lifetimes.
The pump-probe experiment has been commonly
used in laser spectroscopy [3–15], where a pump
laser pulse triggers a reaction in the sample and a
probe pulse monitors optical responses of the sample
after the pump. Laser pump/probe spectroscopy
reveals kinetics of the reactions as well as coherence
between different states of the molecules involved in
photochemical processes, but it cannot obtain struc-
tures of the molecules during the reactions. Conse-
quently, the lack of knowledge in the transient
structures of the molecules greatly hinders fun-
damental understanding of the origins of the func-
tions. Therefore, developing parallel X-ray tech-
niques to determine material structures during light-
induced processes becomes necessary.
Significant progress has been made in recent years
to generate ultrashort X-ray pulses [16–19] for
following atomic movements from the initial
Franck–Condon state to the equilibrated excited

L.X. Chen / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 161–174
163
Fig. 2. Time sequence of the pump-probe XAFS.
2. Experimental methods
fraction of the excited state molecules, the more
likely the transient structure can be determined. The
fraction of the excited state molecules fex is a
function of the laser pulse energy P, the extinction
coefficient of the molecule at the wavelength of the
laser light, ´, the quantum yield Q, as well as the
sample geometry. Eq. (1) describes the relationship
between fex and other experimental factors at a time t
after the pump pulse:
Fig. 2 demonstrates the timing sequence of the
pump-probe XAFS, as well as key factors for the
experiment. A laser pump pulse triggers a photo-
chemical reaction. An X-ray pulse at a certain time
delay from the trigger, t, is used to probe the
structure of the molecules. In addition, an X-ray
detector is synchronously gated with the probe X-ray
pulse to selectively collect signals only at the instant
when the concentration of the transient species of
interest is at its maximum after each excitation pump
pulse. The pump-probe cycle can be repeated until a
sufficient signal/noise ratio of the spectrum is
reached.
Na ? e2kt ? Q
]
fex 5
N
P ? e2kt ? Q
N ? hn
2kt
2´(l)lC 12f Q?e
s
d
ex
f
g
5
? 1 2 10
(1)
where N and Na are the total number of molecules,
and the number of molecules that absorb the light,
respectively; k is the excited state decay time con-
stant, and t is the time after the pump. Q is the
quantum yield of the excitation. P is the pulse energy
of the laser, h is the Planck constant, and n is the
frequency of the laser light. C and l are the con-
centration and the thickness of the sample. Eq. (1)
2.1. Photoexcitation requirements
One of the key requirements for the experiment is
the number of photons from each laser pulse. Be-
cause it is rare to have 100% photoexcitation of the
molecules, XAFS measures a mixture of the ground
and the excited states. Therefore, the larger the

164
L.X. Chen / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 161–174
assumes that the excited state decay follows a single
exponential function and l is small enough so that
the absorption gradient through the pathway of the
light can be neglected. Fig. 3 gives two examples of
calculated Na /N as a function of ´ and C at t50, and
indicates that as ´ gets smaller and C increases, it
becomes more and more difficult to excite a signifi-
cant fraction of the molecules. Using the results from
these calculations, we can estimate fex at t50. The
experimental condition and geometry in the pump-
probe XAFS experiment can be chosen using Eq. (1)
as a guide.
An ideal laser system to serve as the pump for
photo-initiated processes should have a high repeti-
tion rate, a high pulse energy, short pulse width
(FWHM), and a broad tunability in UV and visible
regions. However, the high pulse energy from a laser
inevitably limits the repetition rate of the laser. In
our current experimental set-up, the laser repetition
rate is 1–2 kHz whereas that of the X-ray pulse is of
the order of a few MHz. Consequently, the usable
X-ray flux for the pump-probe XAFS experiment is
reduced by a factor of 1000–2000. This is one of the
challenges that the synchrotron pump-probe XAFS is
facing, which also warrants the need for the high
X-ray flux from the third generation synchrotron
sources.
2.2. Synchronization requirements
Another important factor of the pump-probe
XAFS experiment is the timing structure of the
X-ray pulses and the synchronization of the laser
pump pulse with the X-ray probe pulse. The repeti-
tion rate of the pump-probe cycle is limited by the
recovery time of the reaction and of the detector, as
well as the repetition rate of the pump laser pulses.
These limitations result in a much lower repetition
rate for the pump-probe cycle than that of the X-ray
pulses from a synchrotron source. Consequently, the
signal from a single X-ray pulse that immediately
follows the pump laser pulse must be extracted.
Although a much larger circumference of the storage
ring of the third generation source is potentially
capable of providing a much larger time window for
extracting the signals from a single X-ray pulse, it is
not practical to have only one single electron bunch
circulating in the storage ring. In practice, an
asymmetric fill pattern of the electron bunches is
implemented where the total number of electron
bunches remains large, but one of the electron
bunches is well separated in time from the others
[36]. In order to extract a single X-ray pulse or a
group of pulses, several X-ray shutters have been
designed and constructed for time-resolved studies
[36–38]. The slow speed shutters are made of fast
rotating metal blades and can be used to extract a
part of an X-ray pulse train with sub-millisecond
Fig. 3. The fraction of molecules that absorb light as functions of
(a), (8) when C and ´ are fixed and (b) C when, (8) and ´ are
fixed, at different laser pulse energies.

L.X. Chen / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 161–174
165
opening time. The intermediate speed shutter with a
rotating polygon mirror is capable of extracting X-
ray pulses with one microsecond opening and can be
used to extract a single pulse during the asymmetric
fill operation [38]. The fast speed shutter is a rotating
Si single crystal cube with a pair of slits in the
distance [37], extracting a single X-ray pulse with a
sub-ms opening. With all the designs available,
extracting a single X-ray can be achieved at various
synchrotron sources. Our current experiments, how-
ever, did not use any X-ray shutter. Alternatively, a
gated detection scheme has been implemented and
will be described below.
The synchronization of the laser pulse with the
X-ray pulses from the synchrotron has been accom-
plished by three-stage delays in our experiments
shown in Fig. 4. Under the asymmetric fill operation,
an electron bunch, so-called ‘superbunch’, is in the
middle of a 3.16 ms time window at the Advanced
Photon Source (APS) at Argonne National Labora-
tory. The laser system used in our experiments
consists of a mode-locked, diode-pumped Nd-YLF
laser (Lightwave, LW-131) driven by 1/4 of the
frequency of the RF signal of the storage ring. The
pulses from this seed laser at |88 MHz are thus
intrinsically synchronized to the X-ray pulses. The
seed laser pulses are injected and selectively am-
plified by a Nd-YLF regenerative amplifier laser
(Quantronix). The Q-switch (Medoxx) of the laser is
triggered at 1 kHz by a signal that is delayed and
frequency-divided from the ring revolution signal P0
of 271 kHz. By changing the delay between the
Q-switch of the laser relative to P0, the first delay
determines which superbunch is used to overlap with
one of the laser pulses. Because the 1 kHz laser
pulses are selected from the 88 MHz seed laser pulse
train, the first delay adjustment has a precision
limited to 11.35 ns that separates two adjacent seed
laser pulses. The second delay adjusts the phase
difference between the RF signal and the sinusoidal
Fig. 4. Three-stage synchronization between the laser and the X-ray pulses.

166
L.X. Chen / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 161–174
signal driving the mode-locked laser to 1 ns preci-
sion by a linear delay generator. The third delay
varies the optical path of the laser pulses through an
optical delay line, which can reach a sub-picosecond
precision. However, the time precision for the over-
lap in our present experimental set-up is limited by
the response time of a photodiode detector moni-
toring the laser and the X-ray pulses and is less than
100 ps.
detected photons, (3) detector gating, and (4) timing
requirement. The high photon flux within a single
X-ray pulse from the third generation synchrotron
enables us to conduct the pump-probe XAFS experi-
ment, but some problems of detection also arise due
to such a high flux. For example, the incident X-ray
flux at 11ID-D wiggler beamline of the APS is about
5.531011 photons/s at 9 keV at the sample when the
total storage ring current is 100 mA. During the
asymmetric fill operation, the current in the super-
bunch is 15 mA. Therefore, the incidental photon
flux from the superbunch is 8.331010 photons/s or
3.03105 photons/pulse at 9 keV. In contrast, a
bending magnet beamline at a second generation
synchrotron source provides 109 photons/s at the
sample or about 23 photons/pulse at 8 keV, a factor
of 13,000 smaller than the superbunch at the wiggler
beamline at the APS. The superbunch under the
current hybrid mode operation in the APS is a cluster
of six single pulses with a 2.84-ns separation be-
tween adjacent pulses. If a single pulse within the
six-pulse cluster in the superbunch is concerned,
there is a factor of 2000 increase in the number of
X-ray photons in a single pulse at a wiggler beamline
at APS compared to that at a bending magnet
beamline at a second generation source. An un-
dulator beamline in the APS has a factor of 50–100
higher photon flux than the present state of the
wiggler beamline at 11ID-D at the APS. Therefore, a
pile-up problem for the detector could arise with
such an increase in the photon flux. In order to assess
the problem, an actual number of detected X-ray
fluorescence photons from the sample needs to be
calculated by the equation:
2.3. Detection requirements using a Ge detector
array
In order to obtain an analyzable spectrum with the
pump-probe XAFS, an efficient detector needs to be
selected and the detection scheme needs to be
modified from that of the steady-state XAFS mea-
surements. Many photochemical processes occur in
dilute solutions, so excited state quenching and
exciton–exciton annihilation can be prevented. In
addition, most metalloproteins with interesting
photochemical properties have intrinsically dilute
metal concentration. Thus, the X-ray fluorescence
detection must be used for most of the pump-probe
XAFS experiments. The basic requirements for the
X-ray fluorescence detector for the experiments are
(1) an energy resolution allowing separation of
fluorescence signals of a specific element from those
of elastic scatterings and fluorescence of other
elements with similar atomic numbers, (2) a capa-
bility of synchronized detection with the X-ray probe
pulses, (3) a large solid angle to effectively collect
the signals from a dilute sample, and (4) a fast
enough response time to eliminate interference of
other X-ray pulses to the signals of the gated probe
pulse. Although different X-ray fluorescence detec-
tors are available for dilute samples [39–48], most of
these detectors have not been used for the pump-
probe XAFS. A multi-element Ge solid state detector
(Canberra) [44], which is known to fulfil the first
three requirements, was used in our experiment.
However, the last requirement for the detector has
not been investigated thoroughly because few prece-
dent experiments have been carried out in such a
manner. Issues in the detector application are ad-
dressed here: (1) effects of multi-photons from a
single X-ray pulse, (2) the optimized number of
mk
]
V
If 5 Ia ?
?h ?
?hdet
(2)
4p
mT
where If is the number of fluorescence photons, Ia is
the number of photons absorbed by the sample, S is
the solid angle covered by the detector, h is the
quantum yield of the fluorescence, mk and mT are the
absorption cross sections for the central absorbing
element and for the entire sample, respectively. hdet
is the detector efficiency defined as the fraction of
the fluorescence photons in the total number of
photons entering the detector. Because there are
33105 photons in one superbunch and the pump-

L.X. Chen / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 161–174
167
probe cycle is currently running at 1 kHz, the
number of incident X-ray photons that will be used is
33108 photons/s. For a 1 mM Ni-containing sample
with 1 mm2 area and 0.5 mm thickness, about 15%
photons will be absorbed, which gives Ia 4.53107
photons/s. The solid angle V/4B is 2.9% for a
nine-element detector with a 9-cm2 detection area
situated 5 cm from the sample. The fluorescence
quantum yield h for Ni is 70%, and the fraction of
Ni absorption from the total absorption is 0.6%. With
all the above factors considered, there would be 5500
X-ray fluorescence photons per second entering the
detector, or 5.5 photons/superbunch.
1.7 and three-photon events are at 0.6 by a nine-
element detector. This implies that only 41% in-
cidental photons can be collected as single photon
events that will contribute to the useful signals for
the XAFS spectrum. However, the detector is highly
nonlinear at ns 53.7, so the maximum allowed single
photon event should be much lower. According to
Fig. 5, the nonlinearity of the detector can be largely
neglected when ns ,3, corresponding to N , 5,
giving a maximum of 33% of incidental photons to
be collected as single photon events.
However, only part of the incidental photons
contributes to the XAFS spectrum. A Ge solid state
detector detects both elastic scattering and fluores-
cence photons. As a commonly encountered problem
in collecting XAFS spectra of dilute samples, the
elastic scattering photons often exceed the fluores-
cence photons by a factor of .100. This situation
makes the solid state Ge detector impossible to use.
The obstacle can be removed by using a combination
of the Soller slits and a Z-1 filter from a Lytle
Detector [49,50]. When a cobalt filter with six
absorption lengths was used in our experiment at Ni
K-edge with 1 mM NiTPP, the majority of the elastic
scattering photons were blocked by this filter–slits
combination, and the pile-up due to the elastic
scattering photons was negligible. Although the
cobalt fluorescence appeared as the result of this
set-up, the fraction of Ni fluorescence photons was
increased to 33% of the total photons entering the
detector, whereas it was virtually invisible without
the filter and the slits. In this case, the detector
efficiency hdet is a product of the fraction maximum
allowed single photon events and the fraction of the
fluorescence photons, which gives 11% in our exam-
ple. Therefore, a maximum number of effective
fluorescence photons detected by a nine-element
detector array is 400 photons/s at 1 kHz repetition
rate. Consequently, the average integration time
required to reach 100,000 counts/point is around 250
s. For an XAFS spectrum with 300 points, the total
data acquisition time is 21 h. This result indicates the
limitation of the multi-element solid state detector.
The detector efficiency is severely reduced due to the
elastic scattering from the sample. The detection in
the pump-probe XAFS is detector-limited rather than
photon flux-limited.
The probability of k photon events for each of the
m elements in the detector array is known to follow a
Poisson distribution assuming that the total number
of incidental photons is N,
(N/m)k
k!
P(k, N, m) 5
3 e2N /m
.
(3)
The number of single photon events for the entire
m-element array is then ns(N, m) 5 mP(1, N, m). Fig.
5 displays numbers of one-, two- and three-photon
events as a function of N for a nine-element detector.
When N59, the single photon event detected is at
the maximum of 3.7, while two-photon events are at
Fig. 5. The number of single-, two- and three-photon events vs.
the number of incidental photons N for a nine-element solid state
Ge detector array calculated by Eq. (3).
The gating of a Ge detector array can be accom-

168
L.X. Chen / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 161–174
plished by gating the amplifiers or the scaler. Based
on our experimental results, a minimum of 1.2 ms
time separation is required to prevent interference
between signals from adjacent X-ray pulses when the
detector is operated with a 0.25 ms shaping time.
Such a time separation requirement could be reduced
when a 0.125 ms shaping time is used.
Recently, detectors for the dilute samples have
been constructed with an angle-tuned multilayer
coated Si crystal array [41] or with a bent Si single
crystal [51]. These alternative detectors show a great
potential in enhancing the detector efficiency by
eliminating the elastic scattering photons. However,
the applications of these detectors on the pump-probe
XAFS remain to be further investigated.
3. A recent pump-probe XAFS study
Using the experimental set-up described above, we
carried out a pump-probe XAFS experiment to
capture the structure of a reaction intermediate with a
28-ns lifetime following the photodissociation of
NiTPP-L2 (L5piperidine) in solution (Fig. 6) at the
11ID-D wiggler beamline at APS. The photo-in-
duced axial ligand dissociation/association processes
of heme-like porphyrin derivatives are examples of
how the coordination geometry of a metal ion can be
altered by the electronic structure of its macrocycle
ligands, and serves as a model for oxygen transporta-
tion, catalysis, and photoinduced redox reactions.
According to previous studies [52–54] (Fig. 6),
Fig. 6. Photodissociation/photoassociation reaction of NiTPP. The reaction is reversible and both photodissociation/association can be
triggered by the laser excitation.

L.X. Chen / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 161–174
169
the 3d8 electrons of Ni(II) in a distorted square-
planar NiTPP molecule is mostly at a low spin
(3d )2 (the superscript indicates the electron oc-
2
z
cupancy of the molecular orbital) state in a weakly
coordinating and non-coordinating solvent, but are in
a high spin (3dx 2, 3d ) state with an octahedral
2
2
2y
z
coordination geometry binding two axial ligands in
strongly coordinating solvents, such as piperidine.
Photoinduced electronic p–p* transitions of the
porphyrin macrocycle in NiTPP initiate an electronic
configuration change from (3d )2 to 1(3dx
,
2
2
2
z
2y
3d ) (superscript on the left of parenthesis indicates
2
z
the electron spin state), then an inter-system crossing
from 1(3dx 2, 3d ) to 3(3dx 2, 3d ), which
2
2
2
2
2y
z
2y
z
favors octahedral coordination geometry, leading to
the photoassociation of the two axial ligands in
weakly coordinating solvents. Conversely, similar
p–p* excitation of NiTPP-L2 (L5axial ligand) is
followed by a cascade decay of upper excited states,
leading to a transition from 3(3dx 2, 3d ) to
2
2
2y
z
3(3d )2 and an inter-system crossing to a low spin
2
z
1
singlet (3d )2 state which is stabilized by a square
2
z
planar coordination geometry, causing the dissocia-
tion of the two axial ligands.
Fig. 7. The photodissociation kinetics measured by optical pump-
probe transient absorption. The inset is a part of the optical
absorption spectra of NiTPP and doubly ligated NiTPP-L2 (L5
piperidine, C5H5N) The kinetics were monitored at the charac-
teristic absorption peaks for the molecules. At 530 nm, the growth
of photodissociated NiTPP is monitored, and at 560 and 600 nm,
the depletion of NiTPP-(piperidine)2 is monitored. The optical
transient absorption result at 530 nm suggests that 28% NiTPP-
(piperidine)2 was photodissociated.
The excited state dynamics for NiTPP have been
studied extensively by ultrafast time-resolved optical
transient absorption and Raman spectroscopies
[52,55,56]. Although the processes of the excited
state interconversion in NiTPP-L2 are on time scales
of femto- to picoseconds [52,55], the intermediate
from the photodissociation has a lifetime of over 25
ns (Fig. 7) in a neat piperidine solution before it
recombines with two piperidine ligands and returns
to the 3(3dx 2, 3d ) ground state. Fig. 8 shows
the ground triplet state, T0, by recombining with two
piperidine ligands to form an octahedrally coordi-
nated NiTPP-L2. The transient structures involved in
this recombination are not known. A penta-coordi-
nated NiTPP-L molecule could be involved in the
recombination, the square-planar NiTPP could return
to T0 via a concerted axial chelation with two
piperidine molecules, or an intermediate square
pyramid NiTPP-L structure could be important.
Because it is difficult to identify uniquely the optical
absorption spectrum for each possible intermediate
involved in photodissociation, transient optical ab-
sorption spectra of NiTPP in coordinating solvent
alone are not sufficient to identify the intermediates,
2
2
2y
z
possible pathways for the recovery of the photo-
dissociated NiTPP back to the ground state NiTPP-
L2. Starting with NiTPP-L2, a laser pump pulse
induces an electronic transition from the triplet
ground state, T0, to a group of excited triplet states,
T*. These states decay and undergo inter-system
crossing to an excited singlet state, S*. This excited
state subsequently reacts by ejecting two ligands to
produce the square-planer geometry to S0 [52–54]. If
the solvent were non-coordinating or weakly coor-
dinating this reaction sequence would be completed.
However, in a strongly coordinating solvent the
singlet state is unstable and the system will return to

170
L.X. Chen / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 161–174
Fig. 8. Reaction path of photodissociation of NiTPP-L2, (only the nearest neighboring atoms that are chelated with Ni are shown) where the
ground triplet state T0 is photoexcited to upper triplet states T* (representing a group of possible triplet states). T* then undergo intersystem
crossing to S*, which could occur very quickly, and, finally, ejection of two ligands to form S0. This transient species was captured by the
laser pump/X-ray probe XAFS experiment before it returned to the more stable T0 state of NiTTP-L2.
whereas transient XAFS may be used to directly
obtain the coordination geometry of the Ni atoms
and, therefore, identify the intermediates.
using the same laser as the pump and a xenon flash
lamp as the probe. The decay profile may be fit with
a single exponential function A exp(2t/t), where
t528 ns. NiTPP (Aldrich) was dissolved in piper-
idine (Aldrich) to make a 1 mM solution. A multi-
element solid state Ge detector (Canberra) with a
0.25-ms shaping time was used to collect the Ni Ka
fluorescence. The outputs from the detector am-
plifiers were fed to single channel analyzers (SCAs),
which were coupled to two scaler arrays interfaced
with a computer. One of these arrays was gated with
the laser pulses and the other was not gated or was
gated 14.73 ms later (four periods of the storage
ring).
3.1. Specific experimental conditions
The pump-probe XAFS experiment presented in
the following section was conducted at the Ni K-
edge (8.333 keV). The energy of the X-rays was
selected by a cryogenically cooled fixed offset
double crystal Si(220) monochromator on the wig-
gler beamline 11ID-D, BESSRC-CAT at APS. To
achieve complete coverage by the pump laser of the
sample probed by the X-rays, the X-rays were
focused by a Pd toroidal mirror to a beam size of
0.531 mm2 at the sample, which is less than the
laser beam diameter of 2 mm. The experiment was
carried out during the special operating mode when
the asymmetric fill pattern of the storage ring was
implemented. The electron superbunch generates six
X-ray pulses within 14.2 ns (2.84 ns between
pulses), and this six-pulse cluster is separated by
1580 ns on each side from other pulses. The laser
pump pulses are at 351 nm, with an average pulse
energy of 0.8 mJ. The kinetics of the reaction is
measured by a transient optical absorption set-up,
3.2. Results and discussion
The XAFS spectra are displayed in Figs. 9–12 for
the laser pumped NiTPP-L2 along with those of the
ground state NiTPP-L2 and NiTPP. A distinct shoul-
der at 8.340 keV appears in the X-ray absorption near
edge structure (XANES) of the laser pumped sample
(Fig. 9), which resembles a similar feature in the
ground state NiTPP, and is attributed to the 1s→4pz
transition in a square-planar complex [57]. This
feature was not observed when the laser beam was

L.X. Chen / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 161–174
171
Fig. 11. Fourier transformed XAFS spectra weighted by k3.
Fig. 9. Normalized X-ray absorption m(E) at the near edge as a
function of X-ray energy, E. The shoulder at 8.340 keV for the
laser-pumped sample is indicated by the arrow. The spectrum for
the laser-pumped sample can be approximated by mixing the
ground state spectra of NiTPP-L2 and NiTPP in a ratio of 7:3.
the spectrum of the starting octahedral NiTPP-L2
(Fig. 10). Analysis of both the XANES and XAFS
spectra suggests that the laser-pumped spectrum can
be best described as a linear combination of 7065%
NiTPP-L2 ground state and 3065% NiTPP-like
intermediate, which is in agreement with our calcula-
tion for the excitation yield of 30% under our
experimental conditions, as well as with the transient
optical absorption measurement where the pump
pulse with the same pulse energy depleted 28% of
the peak absorption by the ground state NiTPP-L2.
blocked nor when the detector was gated at a time
long after the ground state recovery. Therefore, the
appearance of this feature indicates that both axial
ligands are removed immediately after photoexcita-
tion, forming a transient square planar NiTPP. The
rest of the spectrum also exhibits differences from
Fig. 10. Background-removed, Fourier transform-filtered XAFS
spectra c(k) as a function of the wave vector k weighted by k2,
where P(k)~3Nj /(kRj2 ) exp(22F j2k2) sin[2kRj 1 *j(k)] (Nj is the
Fig. 12. The back Fourier transformed spectra for the nearest
coordination number, Rj is the average distance, F is the Debye–
Waller factor, and *j is the phase shift).
neighbors. The squares are the fits with the results shown in Table
1.
j

172
L.X. Chen / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 161–174
Fig. 11 shows the Fourier-transformed XAFS spec-
tra, which indicate that the distances from the Ni ion
to the nearest neighbor N atoms for the transient
species are between those of NiTPP and NiTPP-L2 at
their respective ground states. Table 1 displays the
nearest Ni–N distances obtained from a numerical
analysis of these spectra. Each spectrum can be fit
with a single Ni–N distance. The XAFS spectra for
the nearest neighbors are shown in Fig. 12, along
with the fit for the spectrum for the laser pumped
sample. Table 1 lists the fitting results from the data
analysis, and gives an average Ni–N distance of 1.91
two-distance fitting of the nearest neighbors from the
XAFS spectra. In addition, the pre-edge peak of
1s-to-3d transition is not enhanced, as it would be in
the non-central symmetric geometry of a square-
pyramid NiTPP-L. Therefore, the photoinduced in-
termediate formed during the first 14 ns after the
laser pulse is primarily the square-planar NiTPP with
an average nearest neighbor distance that is almost
identical to the ground state NiTPP. This result rules
out the presence of a substantial amount of transient
NiTPP-L. However, if the yield of NiTPP-L were
less than 10% [54], only 3% of the total number of
molecules, it would not have been properly detected.
Moreover, if the lifetime of NiTPP-L were much
shorter than the 14-ns duration of the superbunch, it
would not have been observed.
These results agree with the previously proposed
reaction path for the photodissociation of NiTPP-L2
with a triplet ground state [54]. Upon excitation to an
excited triplet state the system crosses to a singlet
manifold, ejects two axial ligands, and relaxes to the
singlet state S0 within a few hundred picoseconds.
Eventually it returns to the stable T0 ground state via
a thermally activated, diffusion-controlled process.
There is no indication that the association of the
axial ligands occurs in a step-wise process where
NiTPP-L would be seen as an intermediate.
˚
˚
A for NiTPP, and 2.09 A for NiTPP-L2, slightly
deviated from the crystal structures, respectively
[58]. The discrepancy between the solution and the
crystal structures may arise from the dynamic nature
of the ligation at room temperature in solution [58].
The nearest neighbors for the laser-pumped sample
could be best described by two distances of
˚
˚
2.1060.02 A and 1.9260.02 A with a relative ratio
of 3.2:1. This ratio is only slightly less than an
expected value of 3.5:1 and within the experimental
errors for a 30% yield of the square-planar inter-
mediate. Based on structures of a previously studied
penta-coordinated Ni bacterial chlorophyll that has a
similar coordination environment as porphyrin, the
˚
Ni–N bond length increases from 1.95 to 2.04 A
when the coordination of Ni changes from a square-
planar to a square-pyramid geometry [59]. Thus, if
there were a substantial amount of NiTPP-L present,
the nearest neighbor distances would not resemble as
closely those for NiTPP and NiTPP-L2. Moreover,
the fraction of the NiTPP intermediate obtained from
fitting the XANES would not agree so well with the
4. Conclusion
Our results demonstrate that laser pump/X-ray
probe XAFS using X-rays from a third generation
synchrotron source provides a new and powerful
Table 1
The nearest neighbor distances in NiTPPa
Compound
Coordination
number
Bond length
(A)
Debye–Wallerb
factor (A )
2
˚
˚
NiTPP
4.060.5
6.060.5
5.760.5
1.860.5
1.9160.02
2.0960.02
2.1060.02
1.9260.02
0.003
20.003
0.004
NiTPP-(piperidine)2
Laser pumped NiTPP-(piperidine)2
0.005
a Feff7.0 from University of Washington [60] was used to calculate the reference spectra for NiTPP and NiTPP-(piperidine)2 based on
their crystal structures respectively [61,62]. WinXAS 97 [63] was used for data analysis.
b Relative to the crystal structures.

L.X. Chen / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 161–174
173
[4] U. Neukirch, S.R. Bolton, L.J. Sham, D.S. Chemla, Phys.
Rev. B: Condens. Matter Mater. Phys. 61 (2000) R7835.
[5] M. Dahlbom, T. Minami, V. Chernyak, T. Pullerits, V.
Sundstroem, S. Mukamel, J. Phys. Chem. B 104 (2000)
3976.
[6] J.Y. Bigot,V. Halte, J.C. Merle, A. Daunois, Chem. Phys. 251
(2000) 181.
[7] P. Hamm, M. Lim, W.F. DeGrado, R.M. Hochstrasser, J.
Chem. Phys. 112 (2000) 1907.
approach to monitor transient molecular structures in
disordered media. For many metal-containing photo-
active molecules, one of the most important elec-
tronic transitions that govern reactivity is the metal-
to-ligand-charge-transfer (MLCT) transition, which
involves moving electrons from the metal ion to
ligands, producing a transient change in the oxida-
tion state of the metal ion. This change could be
detected by XAFS spectroscopy if the laser pump/X-
ray probe technique can be applied. Inevitably, such
transient oxidation changes could cause atomic dis-
placements as well, which could also be detected by
XAFS. Although the superbunch used in this experi-
ment had six consecutive pulses within a 14-ns time
span that limited the time resolution of this experi-
ment, a single pulse superbunch is scheduled for
future operation at the Advanced Photon Source and
will allow us to capture intermediate structures on
even shorter time scales.
[8] S. Link, C. Burda, B. Nikoobakht, M.A. El-Sayed, Chem.
Phys. Lett. 315 (1999) 12.
[9] A. Kummrow, M.F. Emde, A. Baltuska, M.S. Pshenichnikov,
D.A. Wiersma, Z. Phys. Chem. (Muenchen) 212 (1999) 153.
[10] S. Savikhin, D.R. Buck, W.S. Struve, R.E. Blankenship, A.S.
Taisova, V.I. Novoderezhkin, Z.G. Fetisova, FEBS Lett. 430
(1998) 323.
[11] R. Monshouwer, A. Baltuska, F. van Mourik, R. van
Grondelle, J. Phys. Chem. A 102 (1998) 4360.
[12] T. Kobayashi, M. Kim, M. Taiji, T. Iwasa, M. Nakagawa, M.
Tsuda, J. Phys. Chem. B 102 (1998) 272.
[13] D.M. Jonas, G.R. Fleming, Ultrafast Processes Chem. Photo-
biol. (1995) 225.
[14] S. Lin, H. Van Amerongen, W.S. Struve, Biochim. Biophys.
Acta 1140 (1992) 6.
[15] W.T. Pollard, S.Y. Lee, R.A. Mathies, J. Chem. Phys. 92
(1990) 4012.
[16] R.W. Schoenlein, S. Chattopadhyay, H.H.W. Chong, T.E.
Glover, P.A. Heimann, C.V. Shank, A.A. Zholents, M.S.
Zolotorev, Science (Washington, DC) 287 (2000) 2237.
[17] I.V. Tomov, D.A. Oulianov, P. Chen, P.M. Rentzepis, J. Phys.
Chem. B 103 (1999) 7081.
[18] C. Rose-Petruck, R. Jimenez, T. Guo, A. Cavalleri, C.W.
Siders, F. Raksi, J.A. Squier, B.C. Walker, K.R. Wilson,
C.P.J. Barty, Nature (Lond.) 398 (1999) 310.
[19] C. Rischel, A. Rousse, I. Uschmann, P.-A. Albouy, J.-P.
Geindre, P. Audebert, J.-C. Gauthier, E. Forster, J.-L. Martin,
A. Antonetti, Nature (Lond.) 390 (1997) 490.
[20] B. Perman, V. Srajer, Z. Ren, T.-Y. Teng, C. Pradervand, M.
Wulff, R. Kort, K. Hellingwerf, K. Moffat, Science
(Washington, DC) 279 (1998) 1946.
[21] B. Winter, J. Gatzke, T. Quast, I. Will, M.T. Wick, A. Liero,
D. Pop, I.V. Hertel, Proc. SPIE-Int. Soc. Opt. Eng. 3451
(1998) 62.
[22] H. Oyanagi, A. Kolobov, K. Tanaka, J. Synchrotron Radiat.
5 (1998) 1001.
[23] D.M. Mills, Phys. Today 37 (1984) 22.
[24] C. Bressler, M. Chergui, P. Pattison, M. Wulff, A. Filipponi,
R. Abela, Proc. SPIE-Int. Soc. Opt. Eng. 3451 (1998) 108.
[25] B.C. Larson, J.Z. Tischler, Oxford Ser. Synchrotron Radiat.
2 (1997) 137.
Acknowledgements
The work is supported by the Division of Chemi-
cal Sciences, Office of Basic Energy Sciences, US
Department of Energy under contract W-31-109-
Eng-38. The author would like to thank all the
collaborators over the past 10 years on the time-
resolved pump-probe XAFS and related projects.
They are Drs. James R. Norris, Farrel W. Lytle,
Marion Thurnauer, Pedro Montano, Michael K.
Bowman, Michael R. Wasielewski, Zhiyu Wang,
¨
Peter L. Lee, Wighard J.H. Jager, David J. Gosztola,
Guy Jennings, Anneli Munkholm, and Jan Hessler.
This work would not be possible without their
support and participation. I would also like to
express my sincere appreciation to all the staff
members at BESSRC-CAT, Advanced Photon
Source for their technical assistance.
[26] D.J. Thiel, P. Livins, E.A. Stern, A. Lewis, Nature (Lond.)
362 (1993) 40.
References
[27] D.M. Mills, A. Lewis, A. Harootunian, J. Huang, B. Smith,
Science (Washington, DC) 223 (1984) 811.
[28] L. Powers, J.L. Sessler, G.L. Woolery, B. Chance, Biochem-
istry 23 (1984) 5519.
[29] B. Chance, R. Fischetti, L. Powers, Biochemistry 22 (1983)
3820.
[1] P.W. Atkins, in: Quanta, Oxford University Press, Oxford,
1991, p. 133.
[2] A.H. Zewail, J. Phys. Chem. A 104 (2000) 5660.
[3] P. Kambhampati, D.H. Son, T.W. Kee, P.F. Barbara, J. Phys.
Chem. A 104 (2000) 10637.

174
L.X. Chen / Journal of Electron Spectroscopy and Related Phenomena 119 (2001) 161–174
[30] L.X. Chen, Z. Wang, J.K. Burdett, P.A. Montano, J.R. Norris,
J. Phys. Chem. 99 (1995) 7958.
[46] M. Marcus, L.S. Powers, A.R. Storm, B.M. Kincaid, B.
Chance, Rev. Sci. Instrum. 51 (1980) 1023.
[31] L.X. Chen, M.K. Bowman, Z. Wang, P.A. Montano, J.R.
Norris, J. Phys. Chem. 98 (1994) 9457.
[47] E.A. Stern, S.M. Heald, Nucl. Instrum. Methods 172 (1980)
397.
[32] L.X. Chen, P.L. Lee, D. Gosztola, W.A. Svec, P.A. Montano,
M.R. Wasielewski, J. Phys. Chem. B 103 (1999) 3270.
[33] H. Wang, G. Peng, L.M. Miller, E.M. Scheuring, S.J. George,
M.R. Chance, S.P. Cramer, J. Am. Chem. Soc. 119 (1997)
4921.
[34] E.M. Scheuring, W. Clavin, M.D. Wirt, L.M. Miller, R.F.
Fischetti, Y. Lu, N. Mahoney, A. Xie, J.-j. Wu, M.R. Chance,
J. Phys. Chem. 100 (1996) 3344.
[35] G.K. Shenoy, P.J. Viccaro, D.M. Mills, Argonne National
Laboratory Report, ANL-88-9, 1988.
[36] D.M. Mills, Rev. Sci. Instrum. 60 (1989) 2338.
[37] A. McPherson, W.-K. Lee, D.M. Mills, Proc. SPIE-Int. Soc.
Opt. Eng. 3451 (1998) 139.
[38] A. McPherson, J. Wang, P.L. Lee, D.M. Mills, J. Synchrotron
Radiat. 7 (2000) 1.
[39] C. Karanfil, Z. Zhong, L.D. Chapman, R. Fischetti, G.B.
Bunker, C.U. Segre, B.A. Bunker, AIP Conf. Proc. 521
(2000) 178.
[40] Z. Zhong, D. Chapman, B. Bunker, G. Bunker, R. Fischetti,
C. Segre, J. Synchrotron Radiat. 6 (1999) 212.
[41] K. Zhang, G. Rosenbaum, G. Bunker, J. Synchrotron Radiat.
5 (1998) 1227.
[42] C. Gauthier, J. Goulon, E. Moguiline, A. Rogalev, P.
Lechner, L. Strueder, C. Fiorini, A. Longoni, M. Sampietro
et al., Nucl. Instrum. Methods Phys. Res. Sect. A 382 (1996)
524.
[48] E.A. Stern, S.M. Heald, Rev. Sci. Instrum. 50 (1979) 1579.
[49] F.W. Lytle, R.B. Greegor, G.H. Via, J.M. Brown, G. Meit-
zner, J. Phys. Colloq. C8 (1986) 149.
[50] F.W. Lytle, R.B. Greegor, D.R. Sandstrom, E.C. Marques, J.
Wong, C.L. Spiro, G.P. Huffman, F.E. Huggins, Nucl.
Instrum. Methods Phys. Res. Sect. A 226 (1984) 542.
[51] G.S. Knapp, M.A. Beno, K.J. Gofron, Proc. SPIE-Int. Soc.
Opt. Eng. 3151 (1997) 300.
[52] D. Kim, C. Kirmaier, D. Holten, Chem. Phys. 75 (1983) 305.
[53] D. Kim, D. Holten, Chem. Phys. Lett. 98 (1983) 584.
[54] J. Rodriguez, D. Holten, J. Chem. Phys. 92 (1990) 5944.
[55] H.S. Eom, S.C. Jeoung, D. Kim, J.-H. Ha, Y.-R. Kim, J.
Phys. Chem. A 101 (1997) 3661.
[56] D. Kim, T. Sprio, J. Am. Chem. Soc. 108 (1986) 2099.
[57] T.A. Smith, J.E. Penner-Hahn, M.A. Berding, S. Doniach,
K.O. Hodgson, J. Am. Chem. Soc. 107 (1985) 5945.
[58] S.L. Jia, J. Zhang, J.G. Ma, J.A. Shelnutt, N.Y. Nelson, C.J.
Medforth, K.M. Smith, in: Book of Abstracts, 217th ACS
National Meeting, Anaheim, CA, March 21–25, INOR,
1999.
[59] L.X. Chen, Z. Wang, G. Hartwich, I. Katheder, H. Scheer, A.
Scherz, P.A. Montano, J.R. Norris, Chem. Phys. Lett. 234
(1995) 437.
[60] J.J. Rehr, J. Mustre de Leon, S.I. Zabinsky, R.C. Alberts, J.
Am. Chem. Soc. 113 (1991) 5135.
[61] S.-L. Jia, W. Jentzen, M. Shang, X.-Z. Song, J.-G. Ma, W.R.
Scheidt, J.A. Shelnutt, Inorg. Chem. 37 (1998) 4402.
[62] A.L. McLean, G.J. Foran, B.J. Kennedy, P. Turner, T.W.
Hambley, Aust. J. Chem. 49 (1996) 1273.
[43] L.R. Furenlid, H.W. Kraner, L.C. Rogers, S.P. Cramer, D.
Stephani, R.H. Beuttenmuller, J. Beren, Nucl. Instrum.
Methods Phys. Res. Sect. A A319 (1992) 408.
[44] S.P. Cramer, O. Tench, M. Yocum, G.N. George, Nucl.
Instrum. Methods Phys. Res. Sect. A A266 (1988) 586.
[45] S.S. Hasnain, P.D. Quinn, G.P. Diakun, E.M. Wardell, C.D.
Garner, J. Phys. E 17 (1984) 40.
[63] T. Ressler, J. Synchrotron Radiat. 5 (1998) 118.