Vous êtes sur la page 1sur 10

The Rigaku Journal

Vol. 17/ number 1/ 2000

CONTRIBUTED PAPERS

APPROACHING REAL X-RAY OPTICS


GERHARD HILDEBRANDT AND HANS BRADACZEK*
Research Center of EFG INTERNATIONAL Dueppelstr. 13, 14163 Berlin/Germany; Email: efg-xray-berlin@compuserve.com

1. Introduction
About five years ago we celebrated the 150th
anniversary of Wilhelm Conrad R`ntgen's birthday
(1845, March 27) together with the 100th anniversary
of his famous discovery of a new kind of rays"
(Whrzburg, Germany, Nov. 1895). Between 1895 and
1897 R` ntgen performed a large number of experiments with his "X-Strahlen" in order to find out their
properties and, if possible, to disclose their nature. He
summarized the results in his famous three
"Mittheilungen" (communications) in the "Minutes of
the Wh rzburg Society for Physics and Medicine"
not just easily accessible, but later on reprinted in the
"Annalen der Physik und Chemie" [1].
The most remarkable property of the new rays,
their ability to traverse many materials with small
absorption, found immediately medical application in
surgery. The absorption coefficient seemed to depend
on the density of the material, so R`ntgen concluded
that the rays spread out in a medium containing small
absorbing particles. Consequently the arrangement of
those particles should possibly influence the absorption, for instance in a crystal: irradiation parallel or
vertical to its axis should result in differently strong
absorption. R` ntgen's experiments with calcite and
quartz crystals, however, did not reveal such an
interaction-but had he used a narrower incident beam
and much longer exposure times he could have been
the discoverer of X-ray diffraction already in 1895!
Instead he wrote: "Frequently I searched for diffraction effects, but unfortunately, maybe due to their
week intensity, without success".
In further experiments R` ntgen tried to deflect the
X-rays using prisms, but he could not detect a change
of their directions, so he concluded: it is impossible to
concentrate X-rays with lenses.
All in all R`ntgen discovered and described a lot
of properties of his rays, but must finally confess to be
unable to answer the fundamental question about their

nature. He found some of their properties to be similar


to light (perhaps longitudinally oscillating?), some
others to be not. So this problem had to wait for a new
idea to come.
During a discussion with the then 22 years old
student P. P. Ewald, who asked for advice concerning
problems with his thesis, Max von Laue got this idea
about seventeen years later in Munich: If the atomic
distances in space lattices (which were still hypothetical at that time) were about 10-8cm, as mentioned
by Ewald, and if X-rays were waves with wavelengths
of about 10-9cm, as assumed by A. Sommerfeld, then
diffraction of these waves in crystals should be possible. After a moments hesitation W. Friedrich and P.
Knipping, Sommerfeld's and R` ntgen's assistants,
resp., agreed to perform experiments. After initial
failure (due to Laue's advice to search for diffraction
of fluorescent radiation) the success came about
Easter 1912 as the well-known (Laue-) diagram of a
properly adjusted ZnS crystal, as experimental evidence of the wave character of X-rays and of the
existence of space lattices (this was sometimes called
the birth of solid state physics). Laue added the theory
[2] he correct theory in spite of the fact that he got
problems with its application to the experimental
findings.
It was another young man, the 22 years old
lecturer William Lawrence Bragg (son of the
Professor William Henry Bragg) who solved Laue's
problem. Just from the shapes of the Laue spots in
exposures placed at different distances from the
crystal, he concluded that the incident waves could
have been 'reflected' at sheets of atoms in the crystal.
Leaving the Laue (transmission) geometry he
reported, still in the same year (1912), about directly
reflected X-rays from a cleavage face of mica. This
Bragg (reflection) case opened up the possibility to
study X-ray spectra with the help of the famous
'Bragg equation"

* External Member of the Russian Academy of Natural Science.

13

The Rigaku Journal

n = 2d sin

(=X-ray wavelength, d=lattice spacing constant,


n=order of diffraction, =diffraction angle, later on
called Bragg angle B).
From the atom sites of a hypothetical NaCl lattice
(following from assumptions of the British scientists
W. Barlow and W. J. Pope) W. L. Bragg concluded on
the existence of destructive interferences (Laue
missed this point) and confirmed the proposed structure. Taking known data it was now easy to calculate
lattice spacings, and using the Bragg equation the first
precise X-ray wavelengths resulted: X-ray spectroscopy was born, and using precise wavelengths,
crystal structure determination was founded as a
common work of W. H. and W. L. Bragg [3].
Nowadays Bragg reflection can also be realized
using multilayers-stacks of parallel layers of different
electron density. Mostly they are applied as
monochromators with specific properties, as will be
shown below.
Between 1913 and 1917 Ewald then added
important concepts (structure factor, sphere of
reflection: "Ewald sphere") and finished his series of
papers "On the foundation of crystal optics" with part
III "The crystal optics of X-rays" [4]. This masterpiece
of mathematical physics established the dynamical
theory of X-ray diffraction, valid for perfect crystals.
The solid base for structure determination, however,
was still provided by the much simpler Laue-Bragg
kinematical theory, applicable to less or more imperfect crystals. Ewald's crystal optics of X-rays directly
emerged from his work on light-optical double
refraction in orthorhombic crystals (parts I and II of
his series); but the refraction effects of X-rays are
quite different from those in light optics, for a simple
reason: the index of refraction differs in light optics
significantly from unity (e.g. quartz glass: n=1.459),
whereas it equals nearly 1 in the case of X-rays (x in
nx = 1 - x is in the range of a few 10-5). So X-ray
refraction effects are very small, and this explains why
R`ntgen was unable to detect them with prismsbut
nevertheless they have been used nowadays to
construct some kind of lenses, cp. below (Ch. 2).
Obviously we are confronted with the problem
that the directions of X-rays cannot be influenced to
larger amounts just by refraction. Thus we have no
simple means to change divergences of X-ray beams
as we have in light optics-but this would be needed to
improve experimental X-ray techniques and industrial

Vol. 17 No. 1 2000

applications and finally to achieve a universally


usable X-ray optics.
Just one example: In semiconductor industry the
size and density of the elements on the chips is limited
by the wavelength of light. Using an appropriate Xray optic the resolution could theoretically be improved by a factor of 1000.
In certain cases approaches to X-ray techniques
with changed divergences have already been
reported. In the following chapters some of them will
be discussed in less or more detail.
2. Application of Refraction
As has been mentioned above the index of
refraction is a little smaller than 1 in the case of
X-rays. Therefore the biconcave element shown on
the left hand side of Fig. 1 acts as a converging lens in
X-ray optics, contrary to light optics. Since the effect
of one single element is very small one has to
combine many such "lenses", getting the one-dimensionally focusing system shown on the right hand side
which renders possible to allow for focus lengths of a
few meters. As materials low-Z elements are best
suited (Al, Be, B, graphite); but one has to avoid
stronger Bragg reflections and pronounced Compton
diffraction. Due to the relatively strong absorption
useful applications are restricted to experiments with
synchrotron radiation. As an example, B. Lengeler
and co-workers report on experiments with aluminium CRL ("compound refractive lenses") with n=62
elements using 23.5 keV X-rays; the parabolic shape
of the lenses enables to image areas of 0.3 mm in
diameter with a magnification of 12 and a resolution
of about 0.4m [5]. In another experiment the undulator source of ID22 (ESRF Grenoble) was demagnified from 35x700 m2 to 0.98xl4.3 m2, the measured
gain being 177 [5]. In a following paper the complete
theory of these spherical aberration-free imaging
devices is presented in full detail and compared to

Fig. 1. Optical condensing system based on refraction by


the transfer of X-rays from higher density materials (e.g.
Aluminum) to air. Left hand side: single element. Righthand side: Compound refractive lenses stack.

14

experiments. As intensity gains the figures larger


than 100" for aluminium and "larger than 1000" for
beryllium CRUs are given; a lateral resolution in
imaging of 0.3, resp. below 0.1m, is to be expected.
The following main fields of application are
mentioned: microanalysis with m-beams for diffraction, fluorescence, absorption, scattering; absorptionand phase-contrast imaging of opaque objects which
cannot tolerate sample preparation; coherent X-ray
scattering [6]. In the meantime a focus of 2.5x0.5 m2
at an intensity gain of 1100 has been reached using
19.5 keV radiation [7].
3. Application of Total Reflection
During his attempts to develop an X-ray
microscope the German physicist H. Wolter studied
total reflection at parabolic and hyperbolic mirrors
[8]. Whereas he did not achieve his aim, his considerations provided the base for coma-corrected Wolter
telescopes which are in use nowadays in X-ray astronomy with satellites. The optics of such a telescope
consists of concentric and confocally arranged shell
systems of totally reflecting ( about l) smooth
polished mirrors of parabolic shape, followed by
hyperbolic mirrors enabling to shorten the focus distance. As to give an example: The X-ray satellite
ROSAT contained four such systems, providing an
angular resolution of 5" at focus distance 2.4m. The
detection efficiency is small (in three energy regions
around 1.5, 0.75, and 0.25keV only 100 X-ray
photons per hour were counted stemming from a
starburst galaxy); nevertheless ROSAT detected
about 150,000 stellar X-ray sources.
As a further application of total external reflection, X-rays can be guided in hollow channels (tube or
capillaries). Grazing incidence optics with glass capillaries (Fig. 2) have been successfully used for X-ray
work in the fields of materials research, medical
applications, long-wavelengths studies etc. From
[mrad] 31/E [keV] as critical angle of total external reflection for silica glass one gets 3.8 mrad
(0.22) for CuK and 1.8mrad (0.10) for MoK
radiation. Such small grazing-incidence angles are a
restriction to relatively low energies but offer, at the
same time, the advantage of suppressing high-energy
parts of the incident radiation. Especially useful are
polycapillary optics made of hundreds or thousands of
individual fibers each of them containing e.g. 1000
channels with inner diameters of about 10 m. They
can collimate or focus X-rays collected from cones up
to 10 (Fig. 3). For some applications a sealed X-ray

15

Fig. 2. A glass capillary collecting a divergent X-ray


beam by total reflection.

Fig. 3. A multicapillary system used as a collecting


lens for X-rays.

tube in combination with such an optic can replace a


rotating-anode X-ray source.
Simple straight monocapillaries have been used
as X-ray filters [9] and for distance compensation and
partial monochromatization at rotating-anode X-ray
generators [10]. Using tapered glass capillaries D. H.
Bilderback and co-workers obtained X-ray beams 0.1
m in diameter with intensity gains approaching 1000
at 6keV [11]. With polycapillary optics intense
focused beams for microbeam fluorescence analysis
have been gained [12]; further applications in protein
crystallography, using laboratory X-ray sources,
proved to be especially successful with small, weekly
diffracting crystals [13].
Specular reflection was also the underlying
principle for the construction of small X-ray mirrors
which can successfully be used for work in conventional laboratories. Here, contrary to the application in
X-ray astronomy, the demands concerning the quality
of the image are relatively low. U. W. Arndt combined microfocus X-ray tubes with such mirrors of
parabolic or paraboloidal shape [14]. In this case the
angle of collection of incident radiation can be
increased only by increasing the length of the mirror.
Using elliptical or ellipsoidal mirrors, however, one
can vary the collection angle and the cross-fire at the
sample by an appropriate choice of the magnification
and of the length of the major axis of the ellipse. One
critical problem is the preparation of the reflecting
surface: Highly polished silicon with a very thin gold
coating must guarantee for a surface roughness better
than 1 nm [15].
In the meantime a (patented) "Micromirror" is
commercially available [16] (Fig. 4). Having an

The Rigaku Journal

Fig. 4. Micromirror based on total reflection

ellipsoidal profile polished to sub-nanometer roughness, it focuses X-rays from the narrow line focus of a
micro-tube to a point of 0.3 mm size and 1 mrad
divergence in a distance of 600 mm. The 18 Watt
generator is sufficient to deliver the same photon flux
at the probe as would be available from a 5 kW
rotating anode generator under the same geometrical
conditions. This arrangement is especially useful for
the study of small biological crystals with large unit
cell dimensions.
The just described method is severely limited by
the small critical angles for specular reflection: they
can never be larger than 10 mrad for CuK or 5 mrad
for MoK radiation, resp. Larger angles of incidence
are possible for diffracting gratings with lattice
parameter gradients: graded multilayers, or gradient
(delta-) crystals. This will be discussed in the
following chapter.
4. Application of Interference Effects for
Realizing X-ray Optics
4.1. Introduction, Early Work

Various configurations of X-ray monochromators


are in use since many decades. Depending on the
application they are supplied with flat or bent crystals,
enabling to monochromatize or to focus diverging
incident radiation. Mostly silicon or graphite crystals
are in use, but also other materials, which sometimes
are better adapted to that of the sample. For some
years also multiple crystal monochromators and
multilayers are available.
All these monochromator arrangements, however, are based on the diffraction on less or more
perfect single crystals with a lattice plane spacing
which is constant whithin the whole crystal volum.
Therefore they are not suited for higher sophisticated
X-ray optics. The first step towards such a new
direction, the direction to real X-ray optics, was done
as early as in 1981 by R. K. Smither who proposed "a
new method for focusing X-rays and gamma-rays"
which .makes use of bent diffraction crystals in which

Vol. 17 No. 1 2000

the intercrystalline-plane spacing is varied as a function of position in the crystal". To achieve this
variation Smither discussed the use of thermal gradients as well as the variation of the elemental
composition in the crystal. As possible applications
he mentioned a gamma-ray telescope or the production of real images of strong X-ray sources [17].
By applying a thermal gradient to the crystal the
Bragg angle can be made a variable that could be
controlled as a function of position in the crystal. But
only a small effect is to be expected: a T of 200C
gives a d/d= 2.7A10-3 in quartz, delivering only 200
arcsec for , if B=20.
Considerably larger changes in the d spacing can
be achieved in crystals made of two or more
components. If the Sn content varies from 0 to 6% in a
Ni crystal, the change in d spacing would equal a thermal gradient of 1100C/cm (or even 10 4C when
changing from pure Si to pure Ge). The disadvantage
of this method is that it works with a fixed gradient: it
does not have the ability to change the spacing
gradient in order to adjust for different wavelengths.
Many applications require the crystals to be bent.
Smither discusses the special case of the focusing of a
parallel beam of -rays in a gamma-ray telescope
where the surface of the crystals has to be shaped like
a hyperbola (in transmission) or a parabola (in reflection).
An even broader range of applications could be
achieved by using double focusing. Because each of
the just discussed new crystal elements acts like a
cylindrical lens, one crystal focuses a point source to
a line. Adding now a second element rotated by 90,
double focusing occurs and a point source is focused
to a point image. Double focusing can also be
obtained with a single crystal element bent in two
dimensions, cp. Fig. 5. Using double focusing bent
crystals one could possibly "think of many applications for focusing X-rays or -rays that have parallels
in what is done with visible light".
Finally Smither mentions the "superlattice"
structures (multilayers) which had been proposed just
one year ago at that time [18]. They also could be used
as focusing elements, if one would be able to change
the spacing of the superlattice along the surface of
these elements in a controlled way. Superiattices are
preferably applied to longer X-ray wavelengths.
In another paper Variable-metric diffraction
crystals for X-ray optics", written ten years later,
Smither discusses the possibilities of applying appropriately bent graded crystals to improve the

16

Fig. 5. The application of a thermal gradient for double focusing with an especially shaped single crystal element. From [17].

adaptation of synchrotron radiation to double crystal


arrangements, which are often used in precision
measurements [19]. Whereas the vertical opening
angle of the incident synchrotron radiation is relatively large, mostly 20 to 60 arcsec, the angle of acceptance of the crystals is much smaller, typically only 1
to 10 arcsec: a large part of the incident photons is
thrown away unused. After describing the application of bent graded crystals in two-crystal arrangements to generate monochromatic parallel, or
focused, or defocused beams, Smither reports on lab
experiments with conventional Cu-KL III radiation.
Applying thermal gradients of about 50C cm-1 to
quartz crystals in 2240 reflexion, a large increase of
photons per unit bandwidth could be achieved. In the
case of synchrotron radiation this would lead to a
comparable increase in intensity by a factor of nearly
30.
Meanwhile and independent of Smither's work
the present authors had 1987 applied for a patent
(which was granted later on [20]) and published some
related work [21], where they described the principles
of what they called a "Delta Crystal", which is
identical to Smither's graded (or gradient) crystal.
Also a similar application of "lamella crystals" was
already mentioned there.

17

More recently both methods-graded multilayers


as well as gradient crystals-have successfully been
used to realize optical beam paths of hard X-rays. This
will be reported at in the following sections.
4.2. Hard X-ray Optics Realized with Graded Multilayers

Perhaps one of the best examples for the introduction and application of multilayers for the
realization of beam paths similar to those in light
optics is the paper "Parallel-beam coupling into
channel-cut monochromators using curved graded
multilayers" by Schuster and G `bel [22]. Since 1981
[18], synthetic multilayer X-ray Bragg reflectors (Fig.
6) are known to provide controllable band pass
characteristics, combined with high reflectivity and
stability [23, 24]. Applying tailored thicknesses, a
variation of the period with the depth became possible
for the fabrication of "multimirrors" which are able to
increase the angular or energy range of reflection
[25]. But there exist the additional possibility to vary
the layer thickness laterally, opening up further X-ray
optical applications [26, 27]. One of them is the
adaptation of divergent X-rays to channel-cut monochromators.
Schuster and G`bel mention the example of a
Ge(220) monochromator which has a divergence in
the scattering plane of about 12 arcsec for CuK1

The Rigaku Journal

Fig. 6. Multilayer diffraction at angle B according to the


Bragg law n = 2d sin B.

radiation; this extremely small acceptance angle is


responsible for a low through-put in conventional
laboratory equipments-but it could be drastically
improved if the divergent beam from the source could
be converted into a parallel beam. This argument is
valid also for synchrotron beamlines.
Following a formula for the layer period d, which
varies over the length of a parabola, a W/Si multilayer
with 50 layer pairs was fabricated by sequential
sputtering on a (100) silicon wafer substrate, which
finally was contacted to a steel support with an
appropriate parabolic shape. In experiments, the Xray source provided an opening angle of 24arcmin;
therefore, compared to the above mentioned angle of
acceptance of 12arcsec and assuming a multilayer
reflectivity of 50%, the integral intensity could have
been increased by a factor of 60, if the incident
radiation could have been made exactly parallel. The
measurement, however, delivered a gain of only 6,
maybe due to deviations from the ideal parabolic
shape (and/or from the calculated d-dependence of the
lateral layer structure)*).
In a final chapter of [22] the advantages of a twodimensional collimation by two crossed parabolic
mirrors are discussed, e.g. the insensitivity against
sample misalignment.
4.3. X-ray Optics Realized with Graded Crystals
(Delta Crystals)

Another possibility to realize true X-ray optics


has already been discussed in [19, 20, 21]: the
application of crystals with lattice parameter gradients
d. The production of such gradients by applying
thermal gradients T to single crystals results in
relatively small values of d, but offers the advantage
to change d by changing T The growing of multicomponent crystals, on the other hand, delivers
relatively large but stable values of d.

Multicomponent crystals have been grown since


longer times, but in the beginning intended for
purposes other than X-ray applications. As to give
just one example, R. Triboulet et al. studied the
luminescence properties of the complete Cd 1-xZn xTe
alloy series [28]. Using the THM (travelling heater)
method they grew 80mm long crystals at 900C at a
rate of 4mm per day, extending over large parts of
0<x<l. The radial homogeneity was perfect, the
strong longitudinal composition gradient of 1.25%/
mm, however, introduced strains relieved by fractures. Between the cracks up to 15 mm long single
crystals appeared, mainly near the ZnTe-rich side. At
least in the central part 0.4 < xZn< 0.8 a linear change
in the Zn concentration had been observed.
As may be concluded from this example the
growing of multicomponent single crystals with a
lattice gradient d is possible over wide areas of
changing composition. If however an experimental
application requires large values of d the inevitably
arising strong lattice distortions cannot be avoidedat
least at the present state of the art of crystal growing.
Maybe the incorporation of additional atoms able to
release tensions could be advantageous.
In the last years first successes in growing and
applying gradient crystals are reported at. Using the
gradient projection method large BixSb1-x gradient
crystals have been grown by St. Penzel et al. [29].
Over a length of about 20 mm a linear slope of the
composition gradient had been gained. The tailoring
of well defined concentration profiles seems to be
feasible in this system of complete miscibility.
In another experiment (K, Rb)C8H504 crystals
were grown from aqueous solution due to temperature decrease, starting with a seed of pure KC8H504
and pouring pure Rb-biphthalate with constant rate
into the crystallizer [30]. From the large grown
crystals single blocks with sizes of about l0x4x2 mm3
were cut from different growth sectors. At these cuts
X-ray measurements were performed to check the
crystal perfection, the behaviour of the lattice parameter along the growth direction and the reflectivity
of all parts of the blocks.
Using X-ray films it was shown that CuK rays
incident with 0.3divergence were uniformly Braggreflected along 7 mm of crystal surface (reflection
0.10.0), such proving the capability of a delta-crystal
to serve as a "mirror" for X-rays. The crystals,
however, were far from being perfect: a dislocation
density of 1.6x105 cm2 had been observed; welldistinguished reflection peaks indicated a mosaic

*) In the meantime the gain factor was improved from 6 to 16 (C.


Michaelsen et al.: Adv. X- ray Anal., in print).

Vol. 17 No. 1 2000

18

block structure with disorientations up to 70"; the


composition gradient was not constant and growth
rates seemed to depend on local stresses.
Using K1-xRbxC8H504 gradient crystals with
O<x<0.25 from the same crystal grower (S. V.
Moshkin) and a suited theoretical model, observed
asymmetric diffraction profiles could be satisfactorily
described. Therefore the modeling of such profiles
can be efficiently be used for nondestructive gradient
control of gradient crystals [31].
Since Abrosimov and Rossolenko succeeded in
growing large Si1-xGex gradient crystals [32] new
possibilities for the application of gradient crystals in
many fields of X-ray optics have been offered, e.g. for
optical elements in synchrotron radiation instrumentation [33, 34].
In these laterally graded aperiodic crystals
containing up to some atomic % Ge the lattice
parameter changes nearly linearly along the plane of
diffraction. Thus the variation of the Bragg angle of
divergent incident X-rays on the crystal can be
compensated for, opening up the possibility to operate
a monochromator in the whole energy range above 2
keV without a collimating premirror. The resolution is
nearly crystal limited, the reflected spectral intensity
is increased considerably as compared with a conventional Si monochromator.
As a first application Keitel et al. have
successfully used such SiGe crystals as monochromator material for hard X-rays [35]. From a 125 mm
long Si1-xGex crystal (0.02<x<0.07) grown with a
diameter of 35 mm along the [111] direction by means
of the Czochralski technique [36], a 52x33x7 mm3
plate was cut with the 52x7 mm 2 face parallel to
(110) and 33x7 mm2 face parallel to (111). Over a
crystal length of 40 mm a homogenous variation of
the lattice parameter of about 2.5A10 -3 has been
observed ([35] admittedly contains some wrong
figures, the correct value for d/d is about 8 A10-4 [37],
cp. Fig. 7. Therefore d 111=8A10-4AdSi111 = 8A10 -4A5.43/
3 = 2.5A10-3) . This has been measured with a triplecrystal diffractometer using 120 keV synchrotron
radiation diffracted from (111) planes in transmission.
The advantage of using graded monochromator
crystals is demonstrated in experiments with samples
of lower quality, e.g. in studies of high T c materials: in
these cases the FWHM of the samples compared to
that of a perfect monochromator can be larger by one
order of magnitude, such reducing the available
intensity. The FWHM of a graded monochromator
crystal, however, can be adapted to that of the samples

19

Fig. 7. Measurement of the magnitude of the reciprocal


lattice vector G=2/d [D-1] in 15 different positions on the
side face of a Si, Ge, crystal extending over a length of
I=40 mm. The Ge concentration is decreasing with
increasing I (x about 0.03 at I=40 mm). From G2.0004
at I=0 and G2.0020 at I=40mm one gets G0.0016
and G/G=d/d8.10-4 , as mentioned in the text. (after
[35]).

with diffraction properties which allow reflectivities


up to 100%. As a qualitative reason "this is due to the
fact that in a gradient crystal the variation of the lattice
parameter allows different volume elements to
contribute to the total reflectivity without the locally
diffracted intensity having been attenuated by
extinction in deeper layers of the crystal" [35].
In a recent paper by Keitel et al. measurements
using the just described SiGe gradient crystals have
been compared to theory. As a result it turned out that
a reasonable agreement could be gained only if the
creation of new wavefields in crystals with large
strain gradients was taken into account [38]. As is
well-known since the middle of the 1950th years from
experiments with calcite crystals weekly deformed by
a temperature gradient (selected samples of Icelandic
calcite were the prototypes of "perfect" crystals in
those times) wavefields can follow curved beam paths
[39], a concept already confirmed by certain
theoretical arguments [40]. Later on Penning and
Poider as well as Kato expanded the Ewald/v. Laue
dynamical theory on the base of geometrical optics
such as to be applicable to the propagation of X-ray
wavefields in weekly distorted crystals [41, 42]. From
unexpected contrasts in X-ray topographs, however,
The Rigaku Journal

one was forced to conclude on a new effect: in areas of


strong deformations wavefields belonging to one
branch of the Ewald dispersion surface can "jump" to
the other branch such creating new wavefields (propagating in new directions), and theoretical considerations supported this concept [43]. Taking now this
"interbranch scattering" into consideration CJ cile
Malgrange's calculations in [38] were able to explain
the results of the measurements. Again it was
confirmed what was known to all experimentalists in
the field for long times: crystals with a relatively small
content of week deformations, appearing as being
"perfect" in experiments with long-wavelength Xrays, behave the more imperfect the shorter the
wavelengths of X-rays used in measurements. This
will have to be taken into account in future work with
graded crystals.
5. Application of Fresnel Diffraction for
Realizing X-ray Optics
In this last chapter we will glance briefly at recent
work using fresnel lenses. Up to now we have
discussed exclusively methods which established Xray optical effects with radiation of medium or even
high energy ("hard" X-rays). But since the application
of Fresnel lenses enabled X-ray microscopy to be
successfully realized with long-wave ("weak") X-rays
we have to mention this method at least briefly.
Pioneering work in this field has been done e.g. by a
research group at G`ttingen University headed by G.
Schmahl using synchrotron radiation from the Berlin
storage ring BESSY I (which just now is being
replaced by an improved ring BESSY II).
Recent developments made possible by the use of
high resolution zone plates (e.g. in germanium with
outermost zone widths down to 19 nm) have been
described by G. Schneider et al. [44]. In order to
enhance the image contrast a phase contrast method
has been developed, and to improve the stability of the
mostly examined biological objects a cryogenic object
chamber has been tested, allowing for operation
temperatures down to 130 K. In another paper by G.
Schmahl et al. [45] the fabrication techniques of the
diffraction optics are described in more detail.
In its last state of development the "G`ttingen Xray microscope at BESSY" enables the stereoscopic
and tomographic imaging of frozen-hydrated biological objects at temperatures even below 130 K, thus
increasing the radiation hardness by factors from 103
to 104 compared to unfixed wet objects at room
temperature. For more details cp. the BESSY annual

Vol. 17 No. 1 2000

reports and the books by Michette et al. [46], Aristov


and Erko [47], and Thieme et al. [48].
Bonse et al. used hard X-ray phase focusing by
linear Bragg-Fresnel lenses (BFL, cp. the schematic
drawing Fig. 8) to realize optics for submicron
resolution local diagnostics methods [49], following
the experiences of the just mentioned soft X-ray
microscopy methods. Two identical Si-based linear
BFUs (fabricated by electron beam lithography and
reactive ion etching, F=0.25 m for CuK), crossed at
right angles to each other, delivered point images of
points: this so-called Kirkpatrick-Baez configuration
is able to produce extended images from extended
objects. This has been successfully proved by the
authors using CuK radiation from a conventional
sealed-off X-ray tube. The results suggest that a
demagnification to spot sizes 0.5 x 0.5 m2 should be
feasible. Tests of this BFL microprobe at the
Microfocus beamline at the ESRF (Grenoble) were in
preparation at that time (1991).
Shortly later Aristov et al. reported on similar
experiments at beamline 14B of the Photon Factory
(Tsukuba, Japan) [50] at wavelength 1.2 D, again
using linear Si (111) BFLs with the following
geometric dimensions: Zone widths 10.5 m (innermost) to 0.5m (outermost); aperture 220m; structure height 1.37m (producing a phase shift of 1. at
1.2 D). In the experimental arrangement shown in Fig.
9, a focus size (determined by the outermost zone
width) of 20m was measured, corresponding to a
demagnification factor of 1/20; the efficiency reached
nearly 30%. Analytical expressions enabling quanti-

Fig. 8. Schematic drawing of a linear BFL (mostly


fabricated on Si(111)).

20

light optics, could not be found yet. The most


promising tool seems to be the gradient (delta) crystal,
but just this instrument brought the highest problems
for its production.
Crystal growing and bending could be performed
only in special cases, and without being a prophet it
can be predicted that it will take many years to get
suited delta crystals for an universal X-ray optics, and
this will be one of the highest challenges for crystal
growers.
References
Fig. 9. Experimental layout [50]: 1 asymmetric
monochromator; 2 BFL; 3 observation plane.

tative estimations (of intensity, image contrast etc.)


were developed, hints to relevant literature are given.
A linear Si (111) BFL has also been used to
produce a 2 m focus of 11.6 keV radiation at the
Optics Beamline BL10 of the European Synchrotron
Radiation Facility (ESRF). The samples were III-V
heterostructures grown by the MOVPE (metalorganic vapour phase epitaxy) and MOMBE (metalorganic molecular beam epitaxy) methods. It was
shown that the doublecrystal arrangement "BFL +
sample" meets the requirements for the characterization of advanced III-V semiconductor advices. The
results obtained from rocking curve scans can
demonstrate the uniformity of device areas on a
micrometer scale. For more details cp. the publication
by Iberl et al. [51].
6. Conclusion
Since the old days of the detection of X-rays the
idea of a real optics, comparable to that of the visual
light optics, moved upon research attempts in this
field.
Although Max v. Laue could show that X-rays
have the same character of electromagnetic waves as
visual light it took decades to make the first steps in
the direction to a real optics.
The light optics is based mainly on the refraction
effect. But just this effect was not applicable easily for
X-rays because of the totally different values of the
refraction indices.
On the other hand a real X-ray optics would bring
huge advantages for many technical and scientific
applications.
Contrary to the light optics it became necessary to
use special physical effects for different applications:
Refraction, total reflection and diffraction are used to
solve special problems. A general solution, similar to
21

[1] W. C. R`ntgen: Ann. Phys. Chem. (Leipzig) 64 (1898) 1.


[2] W. Friedrich, P. Knipping, M. Laue: Sitzgsber. Bayer.
Akad. Wiss. (1912) 303, 363. Reprinted in: Naturwiss.
10 (1922) 361.
[3] W. H. Bragg, W. L. Bragg: Proc. Roy. Soc. A88 (1913)
428; W. L. Bragg: Ibid. A89 (1913) 248.
[4] P. P. Ewald: Ann. Phys. (Leipzig) 54 (1917) 519.
[5] B. Lengeler, C. G. Schroer, M. Richwin, J. Tijmmler:
Appl. Phys. Letters, 74 (1999) 3924-3926.
[6] B. Lengeler, C. G. Schroer, J. Tilmmier, B. Benner, M.
Richwin, A. Snigirev, I. Snigireva, M. Drakopoulos:
Accepted for J. Synchrotron Rad. (1999).
[7] B. Lengeler: Private communication (1999).
[8] H. Wolter: Ann. Phys. (Leipzig, (6)) 10 (1952) 94-114.
[9] V. A. Arkadiev, A. A. Bzhaumikhov, H.-E. Gorny, N.
Langhoff, J. Schmalz: SPIE 2859 (1996) 220.
[10] J. J. MOller, H.-E. Gorny, J. Schmalz: J. Appi. Cryst. 28
(1995) 853.
[11] D. H. Bilderback. S. A. Hoffmann, D. J. Thiel: Science
263 (1994) 201-203
[12] N. Gao, 1. Yu. Ponomarev, Q. F. Xiao, W. M. Gibson, D.
A. Carpenter: Appl. Phys. Lett. 69(11) (1996) 1529.
[13] P.-W. Li, R.-Ch. Bi: J. Appl. Cryst. 31 (1998) 806.
[14] U. W. Arndt: J. Appl. Cryst. 23 (1990) 161-168.
[15] U. W. Arndt, P. Duncumb, J. V. P. Long, L. Pina, A.
Inneman: J. Appl. Cryst. 31 (1998) 733-741.
[16] Microsource" (Bede Scientific Instruments Ltd.).
[17] R. K. Smither: Rev. Sci. lnstrum. 53(2), (1982) 131-141.
[18] J. H. Underwood, T W. Barbee Jr.: AIP Conf. Proc. 75
(1981) 170.
[19] R. K. Smither, P.B. Fernandez: Rev. Sci. lnstrum. 63(2),
(1992) 1755-1762.
[20] Patentschrift DE 37 02 804 C2, 1994.
[21] H. Bradaczek, G. Hildebrandt, W. Uebach: Abstracts Xii
ECM Moscow, Vol. 1 (1989) 304; H. Bradaczek, G.
Hildebrandt, W. Uebach: Acta Cryst. A49 (1993) C374.
[22] M. Schuster, H. Gbbel: J. Phys. D: Appl. Phys. 28, (i
995) A270-A275.
[23] M. Schuster, L. Mfiller, K.E. Mauser, R. Straub: Thin
Solid Films 157 (1988) 325-336.
[24] M. Schuster: European patent application 0 456 897.

The Rigaku Journal

[25] K. D. Joensen, P. Hoghoj, F. Christensen, P.


Gorenstein, J. Susini, E. Ziegler, A. Freund, J. Wood:
Proc. SPIE 2011 (1993).
[26] G.F. Marshall: Proc. SPIE 563 (1985) 114-134.

[40] G. Hildebrandt: Zs. Kristallogr. 112 (1959) 312-339;


340-361.
[41] P. Penning, D. Poider: Philips Res. Rep. 16 (1961) 419.

[27] U. W. Arndt: J. Appl. Crystallogr. 23 (1990) 161-168.

[43] J. Gronkowski, C. Malgrange: Acta Cryst. A40 (1984)


512-522. F. Balibar, F. N. Chukhovskii, C. Malgrange:
Acta Cryst. A39 (1983) 387-399.
[44] G. Schneider, T. Wilhein, B. Niemann, P. Guttman, T
Schliebe, J. Lehr, H. Aschoff, J. Thieme, D. Rudolph, G.
Schmahl: Proc. SPIE 2516 (1995) 90-101.

[28] R. Triboulet, G. Neu, B. Fotouhi: J. Cryst. Growth 65


(1983)262-269.
[29] St. Penzel, H. Kleesen, W. Neumann: Cryst. Res.
Technol. 32 (1997) 8,1137-1143.
[30] S. V. Moshkin, 0. M. Boldyreva, T. 1. lvanova, M. A.
Kuz'mina, 1. P. Shakhverdova, P. V. Petrashen, R. N.
Kyutt, H. Bradaczek: J. Cryst. Growth 172 (1997) 226230.
[31] 1. A. Kasatkin, T 1. Ivanova: Crystallogr. Repts. 43
(1998) 1015-1019.
[32] N. V. Abrosimov, S. N. Rossolenko: Annual Rept. lnst.
KristallzOchtung 1996, 12489 Berlin, Germany.
[33] A. Erko, F. Sch5fers, W. Gudat, N. V. Abrosimov, S. N.
Rossolenko, V. Alex, S. Groth, W. Schrbder: Nucl.
Instrum. Methods Phys. Res. A374 (1996) 408-412.
[34] A. Erko, F. Schbfers, W. Gudat, K. J. Sawhney, N. V.
Abrosimov, S. N. Rossolenko, V. Alex, S. Groth, S. W.
Schr`der: SPIE 2856 (1996) 110-119.
[35] S. Keitel, C. C. Retsch, T. Niem`lier, J. R. Schneider, N.
V. Abrosimov, S. N. Rossolenko, H. Riemann: Nucl.
Instr. Meth. A414 (1998) 427-430.
[36] N. V. Abrosimov, S. N. Rossolenko, V. Alex, A.
Gerhardt, W. Schr`der: J. Cryst. Growth 166 (1996) 657.
[37] J. R. Schneider: Private communication (1999).
[38] S. Keitel, C. Malgrange, T Niem`ller, J. R. Schneider:
Acta Cryst. A55 (1999) 855-863
[39] G. Borrmann, G. Hildebrandt: Zs. Naturforsch. Ila
(1956) 585-587; Zs. Physik 156b (1959) 189-199.

Vol. 17 No. 1 2000

[42] N. Kato: J. Phys. Soc. Jpn. 18 (1963) 1785-1791.

[45] G. Schmahl, D. Rudolph, G. Schneider, J. Thieme, T.


Schliebe, B. Kaulich, M. Hettwer: Microelectron. Eng.
(Nether].) 32 (1996) 351-367.
[46] A. G. Michette, G. R. Morrison, C. J. Buckley (Eds.): Xray microscopy 111. Springer Series in optical sciences
67. Springer-Verlag Berlin, (1992).
[47] V. V. Aristov, A. 1. Erko (Eds.): X-ray microscopy IV.
Bogorodskii Pechatnik Publ. Cp. Chernogolovky,
(1994).
[48] J. Thieme, G. Schmahl, D. Rudolph, E. Umbach (Eds.):
X-ray
microscopy
and
spectromicroscopy.
Springer-Verlag Berlin, (1998).
[49] U. Bonse, C. Riekel, A. A. Snigirev: Rev. Sci. lnstrum.
63 (1992) 622-624.
[50] V. V. Aristov, Y A. Basov, T. E. Goureev, A. A. Snigirev,
T. Ishikawa, K. Izumi, S. Kikuta: Jpn. J. Appl. Phys. 31
(1992) 2616-2620.
[51] A. lberl, M. Schuster, H. G`bel, A. Meyer, B. Baur, R.
Matz, A. Snigirev, I. Snigireva, A. Freund, B. Lengeler,
H. Heinecke: J. Phys. D: Appl. Phys. 28 (1995) A200A205.

22

Vous aimerez peut-être aussi