Clocking Femtosecond X Rays
A. L. Cavalieri,
1
D. M. Fritz,
1
S. H. Lee,
1
P. H. Bucksbaum,
1
D. A. Reis,
1
J. Rudati,
2
D. M. Mills,
2
P. H. Fuoss,
3
G. B. Stephenson,
3
C. C. Kao,
4
D. P. Siddons,
4
D. P. Lowney,
5
A. G. MacPhee,
5
D. Weinstein,
5
R. W. Falcone,
5
R. Pahl,
6
J. Als-Nielsen,
7
C. Blome,
8
S. Du
¨
sterer,
8
R. Ischebeck,
8
H. Schlarb,
8
H. Schulte-Schrepping,
8
Th. Tschentscher,
8
J. Schneider,
8
O. Hignette,
9
F. Sette,
9
K. Sokolowski-Tinten,
10
H. N. Chapman,
11
R. W. Lee,
11
T. N. Hansen,
12
O. Synnergren,
12
J. Larsson,
12
S. Techert,
13
J. Sheppard,
14
J. S. Wark,
14
M. Bergh,
15
C. Caleman,
15
G. Huldt,
15
D. van der Spoel,
15
N. Timneanu,
15
J. Hajdu,
15
R. A. Akre,
16
E. Bong,
16
P. Emma,
16
P. Krejcik,
16
J. Arthur,
17
S. Brennan,
17
K. J. Gaffney,
17
A. M. Lindenberg,
17
K. Luening,
17
and J. B. Hastings
17
1
FOCUS Center, Departments of Physics and Applied Physics Program, University of Michigan, Ann Arbor, MI 48109, USA
2
Advanced Photon Source, Argonne National Laboratory, Argonne, IL 60439, USA
3
Materials Science Division, Argonne National Laboratory, Argonne, IL 60439, USA
4
National Synchrotron Light Source, Brookhaven National Laboratory, Upton, NY 11973, USA
5
Department of Physics, University of California, Berkeley, CA 94720, USA
6
Consortium for Advanced Radiation Sources, The University of Chicago, Chicago, IL 60637, USA
7
Niels Bohr Institute, Copenhagen University, 2100 Copenhagen Ø, Denmark
8
Deutsches Elektronen-Synchrotron DESY, Notkestrasse 85, 22607 Hamburg, Germany
9
European Synchrotron Radiation Facility, 38043 Grenoble Cedex 9, France
10
Institut fu
¨
r Optik und Quantenelektronik, Friedrich-Schiller-Universita
¨
t Jena, Max-Wien-Platz 1, 07743 Jena, Germany
11
Physics Department, Lawrence Livermore National Laboratory, Livermore, CA 94550, USA
12
Department of Physics, Lund Institute of Technology, P.O. Box 118, S-22100, Lund, Sweden
13
Max Plank Institute for Biophysical Chemistry, Am Faßberg 11, 37077 Go
¨
ttingen, Germany
14
Department of Physics, Clarendon Laboratory, Parks Road, University of Oxford, Oxford OX1, 3PU, United Kingdom
15
Department of Cell and Molecular Biology, Biomedical Centre, Uppsala University, SE-75124 Uppsala, Sweden
16
Stanford Linear Accelerator Center, Menlo Park, CA 94025, USA
17
Stanford Synchrotron Radiation Laboratory/SLAC, Menlo Park, CA 94025, USA
(Received 2 November 2004; published 24 March 2005)
Linear-accelerator-based sources will revolutionize ultrafast x-ray science due to their unprecedented
brightness and short pulse duration. However, time-resolved studies at the resolution of the x-ray pulse
duration are hampered by the inability to precisely synchronize an external laser to the accelerator. At the
Sub-Picosecond Pulse Source at the Stanford Linear-Accelerator Center we solved this problem by
measuring the arrival time of each high energy electron bunch with electro-optic sampling. This
measurement indirectly determined the arrival time of each x-ray pulse relative to an external pump
laser pulse with a time resolution of better than 60 fs rms.
DOI: 10.1103/PhysRevLett.94.114801 PACS numbers: 41.60.Cr, 41.75.Ht, 42.65.Re
Ultrafast x-ray pulses are providing our first view of
subpicosecond atomic motion. New sources based on
high harmonic generation [1,2] and laser-produced plas-
mas [3] as well as femtosecond laser-sliced synchrotron
emission [4] have been demonstrated. These sources pro-
duce x-ray pulses with durations of less than a few hundred
femtoseconds, the time scale of vibrations in solids and
molecules and the making and breaking of chemical bonds.
While these sources provide the time resolution necessary
to study these dynamics, their relatively low brightness
limits their application and often hinders attempted
experiments.
A new generation of linear-accelerator-based x-ray free
electron lasers (XFELs) will be more than 20 orders of
magnitude brighter than laser-plasma-based sources and
have the potential to produce x-ray pulses below one
femtosecond in duration [5]. With x rays from an XFEL,
researchers can expect to image chemistry in real time on
the atomic scale. While these new XFELs will be far
brighter than any other ultrafast x-ray source, their physical
size and complexity introduce new challenges which, if left
unaddressed, will restrict their application. A major ob-
stacle will be the inability to precisely synchronize the
time-dependent process being studied with the x-ray pulse
generated by a large accelerator-based source.
Subpicosecond time-dependent phenomena are typi-
cally studied with pump-probe techniques in which the
dynamics are initiated by an ultrafast laser or laser-driven
source and then probed after a time delay. If these experi-
ments can be self-synchronized, with the pump and probe
having a common laser source, then precise time delays
can be produced using different optical path lengths. The
time resolution is then limited by the overlap of the pump
and probe pulses that can be as short as a fraction of a
PRL 94, 114801 (2005)
PHYSICAL REVIEW LETTERS
week ending
25 MARCH 2005
0031-9007=05=94(11)=114801(4)$23.00 114801-1 2005 The American Physical Society
femtosecond [6]. An XFEL source will usually provide
only one pulse, either pump or probe. Since timing jitter
between the pump and probe will degrade temporal reso-
lution, the level of synchronization between the XFEL and
an external ultrafast laser is a primary concern. The Sub-
Picosecond Pulse Source (SPPS) at the Stanford Linear-
Accelerator Center (SLAC) is a linac-based x-ray source
with synchronization challenges similar to those that will
confront XFEL sources. The electron acceleration and
bunch compression schemes used for SPPS and XFELs
are nearly identical. SPPS is the first source to employ
compressed femtosecond electron bunches to produce fem-
tosecond x-ray pulses [7,8] and is currently the world’s
brightest ultrafast x-ray source.
Synchronization between a linac-based source and an
external laser to within a few picoseconds can be achieved
for short periods of time if the external laser repetition rate
is tuned to a submultiple of the linac radio frequency (rf).
Pulse-to-pulse variation in electron bunch charge and linac
rf noise, however, introduce variations in bunch arrival
time which effectively prohibit absolute synchronization
at the subpicosecond level. Rather than attempting to im-
prove the synchronization of the external laser to the
accelerator we solved the problem by indirectly measuring
the arrival time of each individual x-ray pulse with respect
to a pump laser pulse. Although the time of arrival fluc-
tuates from pulse to pulse, the information can be used to
place measurements in sequence with precise time resolu-
tion. The time resolution of experiments performed at
SPPS then approaches the fundamental limit fixed by the
x-ray pulse duration.
The pulse-by-pulse timing information is obtained using
a noninvasive technique based on electro-optic sampling
(EOS) of the electric field surrounding the ultrarelativistic
electron bunch that produces the x-ray pulse. Bunches of
up to 2 10
10
electrons are precompressed, accelerated
through the 3 km SLAC linear accelerator to an ultrarela-
tivistic energy of E 28:5 GeV, and then compressed
once more before delivery to SPPS. Electron trajectory
simulations show that the electron bunch length at SPPS
is 10 m rms along the direction of motion, corresponding
to 80 fs FWHM. As a consequence of special relativity, the
Coulomb field of the electron bunch is nearly transverse to
its motion. Its magnitude is the vector sum of the fields of
the individual electrons. We estimate a peak electric field
of hundreds of megavolts per meter at a few millimeters
from the beam.
If a crystal is placed adjacent to the electron beam, its
index of refraction will be distorted anisotropically by the
strong electromagnetic fields associated with the ultrarela-
tivistic electrons [9,10]. This transient birefringence or
electro-optic effect is induced without affecting the elec-
tron bunch propagation. Laser light passing through the
crystal at the same time as the transient birefringence will
have its polarization rotated and can be used to probe the
effect [11]. In the limit that the crystal response is elec-
tronic, this transient birefringence is prompt, and the effect
is proportional to the electron bunch current profile.
Detection of the electro-optic signal that is imprinted on
the probe laser polarization is challenging because the
electron bunch is expected to be shorter than 100 fs.
However, if an ultrafast laser pulse is swept across the
EO crystal at an angle to the electron beam trajectory,
time is mapped into space [12]. As shown in Fig. 1, the
top edge of the laser pulse intersects the crystal before the
bottom edge so that a different cross section of the laser
pulse interacts with the electron beam depending on its
time of arrival. The angle of incidence of the laser probe
determines the sweep rate, and consequently the resolution
of the measurement, while the laser beam diameter deter-
mines the duration of the measurement window. A polar-
izer transmits only the portion of the probe laser where the
polarization was altered by the EO effect, producing a
time-dependent image of the electron bunch. The position
of the signal indicates the bunch’s time of arrival, while its
width and amplitude contain information on the temporal
current profile.
The time resolution of the measurement is limited by the
thickness of the electro-optic crystal [13], the pulse length
of the laser, and the spatial resolution of the imaging
system used to monitor the transmitted laser pulse. For
the measurements presented here we used a 200 m thick
h110i ZnTe crystal positioned 5 mm from the electron
τ
γγ
γ
k
k
k
t
0
t,x
T
(a) (b) (c)
e-bunch
laser
crystal
FIG. 1 (color). Cartoon depiction of spatially resolved electro-
optic sampling. The frames represent three instants during the
course of the measurement. The yellow object represents the EO
crystal. The red object represents an ultrafast laser probe pulse
moving from top left to bottom right. The black oval represents
the ultrarelativistic electron bunch moving from left to right
(with electric field lines indicated). In (a), the front of the
electron bunch and the laser pulse interact in the EO crystal.
In (b), the back of the electron bunch interacts with the laser
pulse. In (c) the electron bunch has passed the crystal and its
shape has been imprinted on the laser polarization profile,
mapping time into space. The intensity profiles of the laser
polarization components are plotted. The width of the signal,
, represents a convolution of the electron bunch length, the
crystal EO response, and the laser pulse duration. The centroid of
the signal indicates the relative time of arrival.
PRL 94, 114801 (2005)
PHYSICAL REVIEW LETTERS
week ending
25 MARCH 2005
114801-2
bunch trajectory. The electric field due to the bunch lies in
the plane of the crystal. The 2 mm wide laser beam is
incident at an angle of 45
with respect to the crystal sur-
face normal with an effective single-shot window of 9 ps.
The laser oscillator operates at an actively-stabilized
repetition rate of 102 MHz, a subharmonic of the linear-
accelerator rf signal at SLAC, and is phase-locked to this
reference to better than 200 fs (rms integrated phase jitter
from 1 Hz to 40 MHz); EOS can be used to measure and
compensate for the jitter and drift existing between the
reference rf and the electron bunch arrival time. To achieve
precise timing, the EOS measurement and x-ray pump-
probe experiments share a common laser system. A single
Ti:sapphire oscillator is located in a laboratory next to the
end of the x-ray beam line, approximately 150 m from the
EOS experiment. This large separation presents a unique
challenge for laser transport and timing, which will be
characteristic of future XFEL facilities.
Ultrafast optical pulses from the oscillator are delivered
to the EO crystal via a polarization-preserving single-mode
optical fiber. Short optical pulses are required for EOS;
however, ultrafast pulses broaden rapidly upon propagation
in an optical fiber. For example, a 100 fs optical pulse will
double its length after traveling only a few centimeters due
to spectral dispersion in the glass fiber. We overcame this
material dispersion by first spectrally dispersing the oscil-
lator pulses using a series of gratings and a programmable
spectral phase mask [14]. The required spectral phase
shifts are programmed into the phase mask by a genetic
learning algorithm [15]. The dispersion is opposite to that
of the fiber, so that the pulses are recompressed as they
travel through the fiber to the EOS experiment. Final
compression occurs in glass optical elements just before
the pulses encounter the EO crystal. In our experiments we
delivered a 135 fs FWHM laser pulse to the EO crystal.
A series of EOS images of the electron bunch, taken as a
function of bunch compression, are shown in Fig. 2. These
are the shortest single-shot electron bunch measurements
to date [16]. The width of the EO signal reaches a mini-
mum of 270 fs FWHM, which is a convolution of the
instrument response and the electron bunch duration, and
appears to be instrument resolution limited. These data
show the optimum compression settings and indicate that
EOS could also be used as an electron bunch length diag-
nostic at future XFELs. A combination of a thinner crystal,
a shorter laser pulse, and better imaging would be required
to resolve the predicted 80 fs bunch duration at SPPS.
Twenty consecutive single-shot measurements made at
SPPS are shown in Fig. 3. The centroid of each EO image
can be determined with 30 fs accuracy, and indicates the
time of arrival of the bunch with respect to the laser pulse.
Random fluctuations from shot to shot are evident and
were measured to be 194 fs rms over 100 s of observation.
The laser fiber transport system can introduce additional
sources of jitter between the EOS measurement and the
x-ray pulse time of arrival at an SPPS x-ray-pump-probe
experiment. In order to determine the magnitude of this
additional jitter we compared timing information obtained
using EOS with that from a laser-pump-x-ray-probe study
of ultrafast nonthermal melting of an InSb crystal [17]. A
strong subpicosecond reduction in Bragg diffraction fol-
lowing intense laser excitation allowed for single-shot
measurement of the relative arrival time of the x-ray pulse
and pump laser pulse [18,19]. This independent and direct
measurement of the relative x-ray timing was made in the
x-ray experiment hutch and could be determined to within
50 fs. Simultaneous x-ray and EOS measurements were
collected over 30 s. The comparison is shown in Fig. 4.
While each timing measurement method individually
yields a jitter of approximately 200 fs, the correlation
shows only 60 fs rms jitter between the two techniques.
02468101214
Time (ps)
Intensity (a.u.)
FIG. 2 (color online). A series of electro-optic signals (offset
for display) collected as the electron bunch compressors at
SLAC were changed, changing the bunch length. The minimum
FWHM shown represents a time of 270 fs which is the resolution
limit of the apparatus.
Shot
TIme (ps)
2 6 10 14 18
0
2
4
6
8
(a)
500 0 500
0
10
20
30
40
50
60
70
∆
t
(fs)
Number of Shots
σ
∆t
= 194 fs
(b)
FIG. 3 (color). (a) Twenty consecutive single-shot electron
bunch measurements. The bright band in each column is the
electro-optic signal, its location indicates the time of arrival of
the electron bunch with respect to the laser probe pulse, and its
width corresponds to the electron bunch duration. (b) Normal-
ized arrival time histogram of 1000 consecutive single shots.
PRL 94, 114801 (2005)
PHYSICAL REVIEW LETTERS
week ending
25 MARCH 2005
114801-3
The short-term timing jitter at SPPS is approximately
200 fs. There are also long-term drifts, associated with the
rf reference to which the laser oscillator is phase-locked
and with the fiber laser transport system. Most pump-probe
experiments, which require accumulation of data over
multiple shots, would have their time resolution limited
by these long-term drifts, which can be as great as 30 ps at
SPPS. Electron bunch measurements made using EOS
could be used to identify and compensate for the sources
of drift. Therefore, a continuous set of indirect x-ray pulse
measurements can be collected and used to improve the
time resolution of an SPPS x-ray pump-probe experiment
from 30 ps to 60 fs or less.
Ultrafast research at linear-accelerator-based sources
requires knowledge of the x-ray pulse arrival time with
respect to a pump laser with femtosecond precision. This
Letter introduces the first technique that specifically ad-
dresses this problem. By measuring the electron bunch
timing, sub-100 fs precision in the x-ray arrival time was
demonstrated. The time resolution for experiments at SPPS
has been improved to 60 fs rms, nearing the resolution limit
imposed by the x-ray pulse duration. Relative timing in-
formation from spatially resolved EO measurements could
probably be extended to 5 fs, matching the projected
performance of XFELs into the foreseeable future
Portions of this research were supported by the U.S.
Department of Energy, Office of Basic Energy Science
through direct support for the SPPS and individual inves-
tigators and through the Stanford Synchrotron Radiation
Laboratory, a national user facility operated by Stanford
University. Additional support for the construction of SPPS
was provided by Uppsala University through a grant from
the Swedish Research Council.
[1] C. Spielmann, N. H. Burnett, S. Sartania, R. Koppitsch,
M. Schnurer, C. Kan, M. Lenzner, P. Wobrauschek, and
F. Krausz, Science 278, 661 (1997).
[2] Z. H. Chang, A. Rundquist, H. W. Wang, M. M. Murnane,
and H. C. Kapteyn, Phys. Rev. Lett. 79, 2967 (1997).
[3] C. Rischel, A. Rousse, I. Uschmann, P. A. Albouy, J. P.
Geindre, P. Audebert, J. C. Gauthier, E. Forster, J. L.
Martin, and A. Antonetti, Nature (London) 390, 490
(1997).
[4] R. W. Schoenlein, S. Chattopadhyay, H. H. W. Chong,
T. E. Glover, P. A. Heimann, C. V. Shank, A. A. Zholents,
and M. S. Zolotorev, Science 287, 2237 (2000).
[5] R. F. Service, Science 298, 1356 (2002).
[6] M. Hentschel, R. Kienberger, C. Spielmann, G. A. Reider,
N. Milosevic, T. Brabec, P. Corkum, U. Heinzmann,
M. Drescher, and F. Krausz, Nature (London) 414, 509
(2001).
[7] P. Emma, R. Iverson, P. Krejcik, P. Raimondi, and
J. Safrarek, in Particle Accelerator Conference, Chicago,
Illinois, edited by P. Lucas and S. Webber (IEEE,
New York, 2001).
[8] L. Bentson, P. Emma, and P. Krejcik, in European Particle
Accelerator Conference, Paris, France, 2002, edited by
T. Garvey, J. Le Duff, P. Le Roux, Ch. Petit-Jean-Genaz,
J. Poole, and L. Rivkin (CERN, Geneva, 2002), http://
accelconf.web.cern.ch/AccelConf/e02/default.htm.
[9] M. J. Fitch, A. C. Melissinos, P. L. Colestock, J. P.
Carneiro, H. T. Edwards, and W. H. Hartung, Phys. Rev.
Lett. 87, 034801 (2001).
[10] I. Wilke, A. M. MacLeod, W. A. Gillespie, G. Berden,
G. M. H. Knippels, and A. F. G. van der Meer, Phys. Rev.
Lett. 88, 124801 (2002).
[11] J. A. Valdmanis and G. Mourou, IEEE J. Quantum
Electron. 22, 69 (1986).
[12] J. Shan, A. S. Weling, E. Knoesel, L. Bartels, M. Bonn,
A. Nahata, G. A. Reider, and T. F. Heinz, Opt. Lett. 25, 426
(2000).
[13] Q. Wu and X. C. Zhang, Appl. Phys. Lett. 68, 1604
(1996).
[14] S. H. Lee, A. L. Cavalieri, D. M. Fritz, M. Myaing, and
D. A. Reis, Opt. Lett. 29, 2602 (2004).
[15] T. Baumert, T. Brixner, V. Syfried, M. Strehle, and
G. Gerber, Appl. Phys. B 65, 779 (1997).
[16] G. Berden, S. P. Jamison, A. M. MacLeod, W. A. Gillespie,
B. Redlich, and A. F. G. van der Meer, Phys. Rev. Lett. 93,
114802 (2004).
[17] A. Rousse, C. Rischel, S. Fourmaux, I. Uschmann,
S. Sebban, G. Grillon, P. Balcou, E. Foster, J. P. Geindre,
P. Audebert et al., Nature (London) 410, 65 (2001).
[18] A. M. Lindenberg, J. Larsson, K. Sokolowski-Tinten,
K. Gaffney, C. Blome, O. Synnergren, J. Sheppard,
C. Caleman, A. G. MacPhee, D. Weinstein et al.,
Science (to be published).
[19] O. Synnergren, M. Harbst, T. Missalla, J. Larsson,
G. Katona, R. Neutze, and R. Wouts, Appl. Phys. Lett.
80, 3727 (2002).
0 102030
0.4
0.2
0
0.2
0.4
0.6
Shot
Time (ps)
0.4 0.2 0 0.2 0.4 0.6
0.4
0.2
0
0.2
0.4
0.6
EOS Time (ps)
Xray Time (ps)
(b)
(a)
EOS
Xray
FIG. 4 (color). (a) Shot-by-shot comparison of pulse arrival
times determined from EOS measurements (blue), and from
ultrafast laser-pump-x-ray-probe measurements (red). Both mea-
surement techniques show a shot-to-shot jitter of 200 fs.
(b) Correlation plot of the data shown in (a). The relative jitter
between the two techniques is 60 fs.
PRL 94, 114801 (2005)
PHYSICAL REVIEW LETTERS
week ending
25 MARCH 2005
114801-4
