Anisotropic Compton scattering in LiF using synchrotron radiation
G. Loupias and J. Petiau (*)
Laboratoire de Minéralogie-Cristallographie (**), Université Pierre-et-Marie-Curie, Paris, France
and Laboratoire pour l’Utilisation du Rayonnement Electromagnétique (***), Bât. 209C, Université Paris-Sud, Orsay, France (Reçu le 31 juillet 1979, accepte le 19 novembre 1979)
Résumé.
2014L’utilisation du rayonnement synchrotron a permis la mesure de la distribution des moments élec-
troniques dans un monocristal de LiF pour les directions 100 > et 111 > avec une résolution de 0,15 unité atomique. Cette résolution est environ 4 fois meilleure que celle des mesures existantes en diffusion 03B3. Le spectro- mètre utilise le rayonnement de LURE-DCI, un monochromateur channel-cut, un cristal analyseur focalisant
et un détecteur à localisation spatiale. Les résultats obtenus confirment l’anisotropie importante de la distribution de moments dans LiF. La comparaison avec les anisotropies calculées dans la zone 0 à 2 u.a. montre un très bon accord avec les calculs LCAO de Berggren et al. pour q 0,6 u.a. et avec les calculs H.F. de Euwema et al. pour q > 0,7 u.a.
Abstract.
2014Using synchrotron radiation, the electronic momentum distribution is measured with a 0.15 atomic unit resolution in a LiF single crystal for the 100 > and 111 > directions. This resolution is about four times better than obtained in 03B3 experiments.
The spectrometer uses the LURE-DCI radiation with a channel-cut monochromator, a focusing crystal analyser
and a position sensitive detector. The measurements performed confirm the significant anisotropy of the electron distribution in LiF. The present results are compared with calculated anisotropies in the range 0 to 2 a.u. ; they
agree quite well with the Berggren et al. calculations (LCAO) for q 0.6 a.u. and with the Euwema et al. calcula- tions (H.F.) for q > 0.7 a.u.
Classification
Physics Abstracts
71.25T - 07.85
1. Introduction.
-Measurements of the electron momentum distribution have recently [1] proved to
be a particularly sensitive probe of the behaviour
of the slowly moving electrons. In condensed matter, such electrons are precisely those which are taking
part in bonding and are thus of a particular interest
for the physicist and the chemist.
Measurements of the momentum distribution for the target electrons are principally obtained by two
methods :
-
the spectral analysis of the inelastically scattered
X or y photons, the so-called Compton profile (CP) ;
-
the long-slit angular correlation profile of the
two photons emitted during the annihilation of
positron-electron pairs (ACP).
The experimental profiles usually lead to the total
electron momentum density n(p) integrated in momen-
tum space over planes perpendicular to the scattering
vector k, i.e. :
with
However this relation is only valid if :
-
in the ACP measurement, the effects of the
positive charge (repulsion by ion core, affinity for
defects and vacancies) can be neglected and the position wave-function can be considered as a
constant. These assumptions are usually far from being true and, as a consequence, the interpretation
of the ACP in terms of electron momentum is some-
times difficult. On the other hand, this technique has
hitherto enjoyed a better resolution than usual CP
techniques and is then to be used for Fermi surface (*) Et E.N.S.J.F., Paris.
(**) Laboratoire associe au C.N.R.S. (LA 09).
(***) Laboratoire du C.N.R.S. associ6 a l’Universit6 Paris-Sud.
Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:01980004103026500
266
studies, when high quality single crystals are avai- lable ;
-
in the CP measurements, the Impulse Approxi-
mation [3] can be assumed i.e. the energy transferred to the recoil electron greatly exceeds its binding
energy. This assumption is always true in the case
of a y-ray source whereas in the case of a classical
X-ray tube it may be different ; then a fundamental advantage of a synchrotron radiation source (SRF)
is the possibility of picking out a suitable incident
photon energy.
The first measurements of Compton scattering
with synchrotron radiation were made by M. Cooper
et al. [4] using radiation from the synchrotron NINA,
at three energies in the range 40-80 keV ; they showed
that the count rates were very favourable while the resolution was severely limited by the use of a solid
state detector.
At LURE-DCI the available energies range from 4 keV to 15 keV.
Compton scattering experiments have been per-
formed to check the accuracy of wave-function calculations [1]. To improve the sensitivity of these checks, it is necessary to improve both the statistical accuracy and the resolution of the data. In order to solve the former problem several highly efficient
y-ray Compton spectrometers [5] have been designed
in the last two years. However the momentum reso-
lution is poor (0.55 to 0.4 atomic unit), being limited by the energy resolution of the analysing solid state
detector.
On the other hand, the momentum resolution of the X-ray crystal analysing spectrometers (typically
0.2 a.u.) is actually limited by statistics considerations for two main reasons : first the inelastic cross section is weak due to the severe competition with the photo-
electron absorption (except for low z) and secondly
the detection is usually performed point by point [6-7].
These considerations led us to choose a high flux X-ray source (SRF), a curved crystal analyser and
a position sensitive detector hoping in this way to
improve the momentum resolution while keeping
a satisfactory statistical accuracy.
The measurements of directional Compton pro- files of LiF using this new spectrometer may be justified
for several reasons. First from the theoretical point
of view LiF is a simple system and directional Compton profiles may be derived according to two models :
either by Hartree-Fock calculation using a Gaussian
basis set by Euwema et al. [8] or by a tight binding
calculation using Kunz’s crystal orbitals by Berggren
et al. [9]. Secondly y-ray Compton profiles are avai-
lable [9] and offer a good opportunity for a comparison
of the assumptions and data corrections involved in both X and y-ray techniques.
2. 1’he X-ray spectrometer.
-2.1 APPARATUS
DESCRIPTION. - The SRF photons being mostly horizontally polarized [10] the spectrometer has been
built in the vertical plane. A schematic diagram is
shown in figure I and a detailed set-up in figure 2.
Fig. 1. - Schematic representation of the Compton spectrometer
at LURE-DCI.
The white synchrotron beam is monochromatized by a double reflection in a channel-cut single crystal.
In the present experiments 220 reflections in the
symmetrical Bragg case have been used in a silicon
crystal but, in order to increase the monochromatic beam flux (by a factor roughly equal to four) a ger- manium single crystal has been cut [11] in such a way that the reflections 220 are asymmetric with angles of respectively 10 and 7 degrees between the reflecting planes and the surfaces. The incident wavelength Ao
is selected by use of rotation Rl with a step of two seconds of arc and the whole spectrometer is matched
to the position of the monochromatic beam through
the translation T 1.
The vertical width of the slit F1 has to be limited
in order to improve the resolution (under a width
of 1.5 mm the incident energy resolution is controlled
by the source size). The slit, systems (F2, F3) and the
monochromator and analyser shielding are dictated by background considerations. Moreover in order to save intensity and reduce the background, the mono-
chromator as well as the paths of the incident and analysed beams are included in an evacuated chamber.
The incident flux is monitored in an air ionization chamber I and irradiates the sample S mounted on a goniometer head fixed on a rotation axis R2 so that any hkl > crystallographic axis can be oriented
along the scattering vector k. The value of the scatter- ing angle T is adjusted by the rotation R3. For this
configuration of the spectrometer (Fig. 2) the value of T can be selected at will from 1000 to 1700.
The scattered radiation is energy-analysed by Bragg reflection in a curved crystal A, in the trans- mission geometry. The analysed beam is focused
on the Rowland circle. The coupling of a position
sensitive detector (PSD) with a multichannel analyser
allows all the points of the spectrum to be simul-
taneously collected. The crystal plate A is cut in such
Fig. 2.
-Experimental set-up of the Compton spectrometer at LURE-DCI. M : channel-cut monochromator (Ge or Si). I : ionization chamber. S : sample. A : curved analyser (Si 220). D : position sensitive detector.
’a way that the Bragg reflected beam is normal to
its surface. This is indeed the geometrical condition providing both the best crystal focusing [12] and
the smallest parallax in the 5 mm thick detector of the backgammon type (Jeu de Jacquet) [13]. The rotation R4, actually achieved by a vertical translation of the bracket shown on the right side of figure 2, allows
the analyser adjustment for Bragg reflection of the
Compton scattered wavelength. To reach the best resolution available with this spectrometer configu- ration, the crystal may be bent with a radius of cur-
vature of up to two meters. The PSD is always posi-
tioned normal to the diffracted beam (Fig. 2). All
these motions. are motorized and remote controlled.
Finally and still for background reduction, the geometry has been designed in such a way that the detector should not be in direct view of any point
of the sample.
2.2 COMPARISON BETWEEN THE -PERFORMANCES OF THE PRESENT APPARATUS AND A SELECTION OF COMPTON
SPECTROMETERS USING THE DIFFERENT AVAILABLE TECH- NIQUES.
-All the data presented in table I are
concerned with a 0.2 mm thick aluminum sample
and have been derived from the information given
in each experiment. For y-ray spectrometers, the sample thickness can be increased up to a few milli-
meters (usually 1 mm) with a corresponding increase
in intensity but also in the amount of multiple Compton scattering events which requires careful
corrections. For the DCI spectrometer, the data have been recorded with the germanium mono-
chromator, the incident energy being fixed at 10.5 keV
and the scattering angle at 140°. The conditions of the impulse approximation are satisfied for all the electrons : the energy transfer at the Compton peak
is 380 eV while the electron binding energies are res- pectively 1 559 eV and 72-87 eV for K and L shells
in aluminum.
,The experimental data listed in table I clearly
demonstrate the large gain in resolution obtained with the curved crystal analyser set-up as compared
to the y-ray spectrometers. The resolution function has been evaluated through the width of the fluores-
cence line Kocl of germanium used as a sample.
No serious drop in the statistical accuracy is observed although aluminum is not the most favou- rable case for a low incident energy source. However
an increase of the analysing crystal quality can bring
the resolution down to 0.05 atomic unit without
intensity loss; this X-ray spectrometer will then be
an order of magnitude more sensitive than a y-ray spectrometer with a SSD detector.
2 . 3 DIFFERENT CONFIGURATIONS OF THE APPARATUS.
-