ESRF Newsletter Oct. 2000
Hard X-ray Holography at the ESRF
M. Belakhovsky1, G. Faigel2, S. Marchesini1,
M. Tegze2 and O. Ulrich1
1 -
SP2M / CEA-Grenoble (France)
2 - Research Institute for Solid State
Physics and Optics, Budapest (Hungary)
Hard X-ray holography permits the visualisation of the 3-D arrangement
of atoms at the angstrom level. The development and future prospects of
this new technique are described.
Introduction
The knowledge of atomic and molecular structures is fundamental to physics,
chemistry and biology. From the beginning of this century, much effort has
been put into the development of reliable methods for structure determination.
Diffraction techniques are the most common and the most highly developed
for structural studies. However, not all of the problems can be solved by
diffraction: well-known difficulties are that single crystals cannot be
grown for all substances, and that a full structure determination is difficult
from powder data alone. Sometimes the phase problem causes diffraction to
fail even with single crystals. As for non-periodic systems, like
amorphous materials, quasi crystals etc., detailed structural information
cannot be derived easily. Further difficulties arise with new artificial
materials, which have structures that cannot always be solved by traditional
crystallographic methods. Examples are the study of the local environment
of low-concentration impurities in a sample, the atomic order in buried
interfaces, and the atomic environment of surface adsorbates.
Therefore, it is not surprising that scientists are trying to find new
techniques for structural investigations. Recently, atomic-resolution X-ray
holography has emerged. It is based on the same principles as traditional
holography with light: i.e. a coherent wave (called the reference wave)
illuminates the object and the detector surface. The intensity modulation
caused by the interference between the reference wave and the wave scattered
by the object (called the object wave) is recorded. This interference pattern
contains both the phase and the magnitude information of the object wave.
Therefore, the original wavefront can be reconstructed, giving the 3-D spatial
arrangement of the objects. Although holography is used in many areas of
science and everyday life, atoms in solids could not be imaged until recently.
This is because the resolution is limited by the pixel size of the detector
and the size of source, in addition to the wavelength. While it is relatively
easy to decrease the wavelength to the Angstrom level, giving the possibility
of atomic resolution, it is difficult to reduce the source size or to increase
the detector's resolution. Abraham Szöke pointed out that individual
atoms in a solid could be used as sources of radiation [1]. Based on his idea, experiments were performed,
first using electrons [2] and
later using photons [3] as hologram
forming waves. Moreover, Gog and co-workers showed that the atoms present
in a sample could be used as detectors instead of sources of radiation [4]. In spite of these pioneering
efforts, many problems relating to the experiment and evaluation of results
remained to be solved. The most significant ones were the long measuring
time, the experimental setup for synchrotron studies, background correction,
and the method of reconstruction.
These problems prompted us to start a series of experiments at the ESRF.
Our aim was to develop a prototype experimental setup for synchrotron holographic
studies and at the same time to find reliable methods for background correction
and reconstruction. In this article we would like to describe the most important
results from these experiments.
For clarity, first the principle of the
"inside source and detector" holography is given. In our experiments
the electronic system of the atoms is used as sources or detectors of X-ray
radiation. Figure 1a shows the
formation of a hologram in the case where the atoms act as point sources.
A central atom is excited by an external source. In the de-excitation process,
a fluorescent photon is emitted in the form of a spherical wave. This wave
can reach the detector surface directly or after scattering by the neighbouring
atoms. The two waves interfere producing an intensity modulation which is
measured on a spherical surface surrounding the sample. The intensity modulation
contains the hologram. In the other case i.e. when atoms are used as "point
detectors" the external wave is incident to the sample in the form
of a plane wave (Figure 1b). This can reach the central atom directly or
by scattering on the neighbouring atoms. The two waves interfere at the
detector atom (central atom) and the resulting field excites the electronic
system. The probability of excitation is proportional to the strength of
the field. Changing the direction of the incident radiation changes the
phase relation between the direct and scattered waves resulting in oscillations
of the fluorescent intensity. These oscillations contain the holographic
information [5].
Experimental developments
It is clear from the above description, that the two types of measurements
require a similar experimental setup. The main difference is that in the
inside source case the fluorescent radiation has to be measured as a function
of the detector position in relation to the sample. While using the atoms
as "point detectors" the direction of the incident beam has to
be varied and the external detector has to collect the fluorescent photons
from the full solid angle about the sample. The first step in our work was
to develop the proper experimental setup for synchrotron measurements. It
was constructed in a way that both inside "source and detector"
holographic experiments could be done without remounting the sample. One
of the critical components of the setup is the detection system. This should
satisfy three conditions: first it has to discriminate the fluorescent photons
from a relatively high background, second its angular acceptance has to
be variable from the 1*1 degree2 to the
full solid angle (or as large a solid angle as possible), third it has to
handle high count rates without appreciable dead time. The mechanics is
another crucial part of the experimental setup. Although angular positioning
and reproducibility do not have to be extremely precise, it does have to
be fast - capable of 10 turns/second for one of the axes. For the two other
axes rotation speed can be slower, about 2 degrees/second, but they both
have to carry relatively heavy loads (3 kg). The third part of the experimental
setup which needs attention is the data acquisition system. During the fast
motion one has to read several counters in quick succession (every 200 microseconds).
We developed a setup which satisfies the above conditions, improving
the system after every experimental run. Here we will not describe all the
steps but instead present the latest version of the setup which uses the
full power of synchrotron undulator radiation. The experiments were done
at beamlines ID32, ID18 and ID22.
The experimental setup is shown in Figure
2, and a description of its characteristics follows. The optics are
the least complicated part. A given harmonic of an undulator is used without
monochromatisation since a small bandwidth is unnecessary. However, a mirror
and an absorber are needed in order to single out a given harmonic
the so-called "pink" beam which then passes through a thin
foil, which scatters part of the incident beam. The scattered radiation
is used for incident beam monitoring. The mechanical system comprises two
coaxial vertical rotation stages (
, ') coupled in such
a way that the lower goniometer () carries the upper one ('). The upper goniometer in turn holds another smaller one with a horizontal
axis (
). The sample is fixed to this horizontal rotation stage
so that its flat surface is perpendicular to this axis when illuminated
by the X-ray beam. The detector is mounted on the lower goniometer. The
third part of the experimental setup is the detector assembly. Since the
count rate is very high, it could not be handled by single photon counters.
Therefore, we used a Si diode in the current mode. However, this does not
have energy resolution, which is necessary to suppress the unwanted radiation,
so the energy analysis was done before the detector by a doubly focusing
pyrolithic graphite monochromator (Figure 2 [6]). This arrangement resulted in a large loss of detection
solid angle compared to the laboratory measurements. Fortunately, this was
more than compensated for by the high flux of synchrotron radiation. Typical
estimated count rates were in the range of 1010
counts per second. Taking into account the statistical noise only, this
would result in measuring times of a few seconds. However, the mechanical
motion of the sample-detector system is not fast enough to scan a hemisphere
in this short time. Therefore, the collection of a holographic data set
at a single energy takes a few minutes.
Holographic images
In the following part we would like to present a few examples of holographic
imaging. Useful information can be derived from the holograms without any
back-transformation. Since we use the atoms as point sources or detectors
in the hologram forming process, the hologram shows the local symmetry of
their environments. This is illustrated in Figure 3. The pictures show the recorded fluorescence intensities,
projected on the sample surface of the upper hemisphere, after absorption
correction. What is really seen is the standing wave patterns (or Kossel
lines) rather than the holograms themselves, since the holographic oscillations
are of much smaller amplitude. The pictures a, b and c were taken from samples
having a three, four and five-fold symmetry axis perpendicular to their
surface, respectively from an NiO[111] single crystal, an epitaxial FePt
L10 film and an AlPdMn quasi-crystal. One
can find other off-perpendicular symmetry points on these holograms from
which the full local symmetry of the site can be deduced [7].
Of course, what we are finally interested in is the 3-D arrangement of
atoms in real space. The next two figures show an example of this. The holograms
of a CoO[111] sample are shown at four different energies (Figure 4).
The reconstructed image of the Co atoms can
be seen in Figure 5 [7]. In the
evaluation process all four holograms were used to build a single high-resolution
atomic structure. We would like to point out that in this case the resolution
is isotropic in contrast to all previous measurements, and its value (0.5
Å) is near to the diffraction limit.
Future trends
In spite of the substantial progress in the experimental and evaluation
methods, there are many problems which need to be solved before the widespread
application of this technique. Let us start with the easiest one to solve:
the collection time of a hologram, which at a single energy is between 5
and 30 minutes. This does not seem too long, especially when compared with
the duration of the first demonstration experiment (two months [3]). However,
it would be advantageous to decrease the measuring time to 1 minute or below.
The reason for this is that many artifacts inherent to holographic reconstruction
can be eliminated by measuring the same sample at several (in the order
of 10) energies [4,5,7]. In practical terms this would result in measuring
a full data set for holographic reconstruction in 10 minutes instead of
a few hours. The second problem is that we are unable to image light atoms.
In all experiments so far, only relatively heavy atoms (from iron up) could
be imaged. However, in many applications especially in the case of biological
samples, it would be crucial to see O, C, N etc. The third problem that
hinders wider application of holography is the form of the sample. Presently
single crystal samples with a large (5*5 mm2)
flat surface can be measured. It would be very useful to extend the capabilities
of holographic imaging to small (~103
mm3) arbitrarily shaped crystallites.
Of course there are many other areas where holography could be improved,
but we stop here after mentioning the three most important ones. We have
continued to work on the above mentioned problems, and have promising results,
especially on the measuring time, on the reconstructed cluster size and
the imaging of light atoms [8] and
on the first application to thin films.
References
[1] A. Szöke, in Short Wavelength Coherent Radiation: Generation
and Applications, AIP Conference Proc., ed. D.T. Attwood and J. Boker (New
York) 147, 361 (1986).
[2] G.R. Harp, D.K. Saldin, B.P. Tonner, Phys. Rev. Lett., 65, 1012 (1990).
[3] M. Tegze, G. Faigel, Nature, 380, 49 (1996).
[4] T. Gog et al., Phys. Rev. Lett., 76, 3132 (1996).
[5] For a more detailed discussion of atomic resolution X-ray holography
see: G. Faigel, M. Tegze, Rep. Prog. Phys., 62, 355-392 (1999).
[6] S. Marchesini, M. Belakhovsky, A.K. Freund, in Crystal and Multilayer
Optics, SPIE proc., 3448, 224 (1998).
[7] M. Tegze, G. Faigel, S. Marchesini, M. Belakhovsky, A.I. Chumakov,
Phys. Rev. Lett., 82, 4847 (1999).
[8] M. Tegze, G. Faigel, S. Marchesini, M. Belakhovsky, O. Ulrich, Nature,
in print.
Acknowledgements
We would like to thank the beamline
staff at ID32, ID18, and ID22 for their assistance, especially A.I. Chumakov,
L. Ortega and A. Simionovici. S.M. acknowledges the E.U. Marie Curie Fellowship
under grant n° ERBFMBICT961366. This work was also supported by Balaton
P.A.I. n° 98024 and OTKA T022041.