Professional Documents
Culture Documents
Ch7Spectroscopic Approaches
Ch7Spectroscopic Approaches
Chapter 7
Spectroscopic Approaches
Esther Belin-Ferré
Universite´ Pierre et Marie Curie – Centre National de la Recherche Scientifique, Paris, France
Figure 7 Mapping of the Fermi surface of Pt(111) (left side) from PES measurements at room temperature with an incident
radiation of energy 21.2 eV and its comparison with theoretical calculations (right side). Black dots of the calculation are to be
compared with high intensities in the experiment
Spectroscopic Approaches 141
or
Figure 9 Sketch of the principle of X-ray emission. Panel a
IðhnÞ / jMf!i j2 NðEÞ*Lðn, l, j Þ ð6Þ represents an X-ray transition between two core-levels. Panel
b corresponds to an X-ray emission involving the occupied
for two levels, Lðn, l, j Þ and Lðn0 , l 0 , j 0 Þ, equation (5), band
or for one level and the OB (equation (6)) participating
in the transition, respectively. In equation (6), NðEÞ
denotes the probed OB densities of states. Mf!i is the
matrix element of the transition probability; it is probability differs from 0, which is achieved provided
constant or varies slowly with energy depending on the transitions fulfil dipolar selection rules, namely:
jh f jrj i ij where i and f are the wave functions of l ¼ 1, j ¼ 0, 1 ð7Þ
the initial and final states. Therefore, in equation (5),
the initial and final states are highly localized; in As a consequence, if a core hole with symmetry l is
contrast, in equation (6), i is a localized state whereas created in an element A, alone or in a compound, the
f describes a hole in the OB. Spatial overlapping local density-of-states with l 1 symmetry around this
between the initial and final wave functions is required element, denoted by NðEÞl1 will be probed by the XES
for a transition to take place, and this requirement technique, thus giving also a partial character to XES.
implies that only the amplitude of f in the core hole Note that (i) transitions with j 0 ¼ 0 ! j ¼ 0 are not
region will give a significant contribution to the allowed; (ii) transitions to l þ 1 states are favoured
spectrum. Accordingly, XES has a local character. In with respect to those to l 1, therefore, from a p hole,
addition, it is necessary that the radiative transition states with d character will be probed with a higher
142 Bonding and Stability
probability than states with s character. Consequently,
comparison of the experimental XES spectra with
partial densities of states calculations is possible. More
details can be found in many textbooks, as for instance
Agarwal (1979), or in Bonnelle (1987).
The shape of the distribution for a transition
involving two inner levels is a Lorentzian curve
whose FWHM is the sum of the FWHMs correspond-
ing to the two levels involved in the transition. The
shape of the emitted intensity distribution to which the
OB contributes results from the convolution of the
probed partial local DOS and the Lorentzian distribu-
tion Lðn, l, j ).
The initial hole can be created either with impinging
electrons or with photons. Irradiation by electrons
easily produces multiple interactions; hence, satellite
emissions due to the X-ray transition in the multiple-
ionized or excited system may also be observed in the
spectra. The XES technique is straightforward, directly Figure 10 Mg 3p distribution of states in the pure metal
applicable to any kind of material: crystalline, amor- (lower curve) and in NiMg2H4 (upper curve) on the X-ray
phous, insulator, bulk, powder, etc. In addition, transition energy scale
according to the conditions of the experimental set-
up it can inform the investigator either with reference
to the bulk sample or to a thin layer just below its a picture of the OB and have insight into the electronic
surface. OB of deep interfaces in solids may also be interactions that are present, further adjustment of the
investigated with XES induced by electrons by tuning various partial contributions on the energy scale is
the energy of the incident beam so that it reaches the necessary. This can be realized on the binding-energy
depth of the interface in question (Fargues et al., 1985; scale but requires knowing the Fermi energy EF for
Szasz and Kojnok, 1985). The total energy resolution each individual X-ray transition energy scale. To do
in an experiment results from the natural width Eðn, so, it is necessary to determine the binding energy with
l, j ) of the inner level Lðn, l, j ) and the instrumental respect to EF of the inner level participating in each X-
function. Eðn, l, j ) varies from a few meV up to ray transition; this can be achieved thanks to
several eV according to the atomic number and complementary X-ray and (or) inner-level PES mea-
considered level (Krause and Oliver, 1979); and it is surements. An example is shown in Figure 11 for
generally the most important parameter in choosing Al7 Cu2 Fe. It presents the curves corresponding to the
the appropriate transitions to investigate. Despite the Al 3p, Al 3s--d, Cu 3d--4s and Fe 3d--4s contributions
fact that no absolute DOS values are obtained, it to the OB, respectively, as adjusted to the binding
makes sense to compare the DOS of a given element in energy scale from the measurements of Al 2p3=2 , Cu
various materials; because, according to equation (6), 2p3=2 , Fe 2p3=2 inner levels and the X-ray transition Al
any change in the shape and energy of the spectral 2p3=2 ! Al 1s. Each curve is normalized to its own
distribution will give direct insight into modifications maximum intensity. Note that as transition probabil-
of the corresponding N(E). An example is given in ities strongly favour d states, no actual view of the Cu
Figure 10, which shows the Mg 3p distribution in the and Fe 4s states is achieved. The overlap between the
pure metal and in Mg2 CoH5 (Belin et al., 1987). The peaks for Fe 3d and Al 3p curves denotes a mixing of
feature in the low transition energy range, not present the corresponding states and therefore illustrates some
in the Mg 3p band in the lower curve, corresponds to hybridized, covalent character of the Al--Fe bond. The
additional states due to Mg-H bonding in the maximum of the Cu 3d peak is almost at a relative
hydrogenated intermetallic. minimum of the Al 3s--d distribution curve. Here, this
A relevant characteristic of XES is that the can be interpreted as a Fano-like interaction (Ter-
contributions of the various components to the OB akura, 1977) between the localized d states and the
of a solid can be investigated separately, each one extended Al states which splits them into two bonding
obtained in its own transition energy scale. To achieve and non-bonding parts, located on each side of the Cu
Spectroscopic Approaches 143
3d maximum (Belin et al., 1992). There is a decrease of
intensity at EF of the various counterparts as
compared to pure metallic elements, which indicates
the opening of a small pseudo-gap at EF in agreement
with the fact that this alloy is a Hume-Rothery
intermetallic.
Comparison between experimental spectra and
partial DOSs is shown in Figure 12 (Trambly de
Laissardière et al., 1995). The agreement is very good
for the Cu and Fe states, but for Al 3p and 3d states
there is a slight discrepancy in the energy range 4 eV
below EF which is still a matter of investigation by
Figure 11 Al 3s-d (dots), Al 3p (solid line), Cu 3d (stars) and theoreticians (Papaconstantopoulos, 1997). Note that
Fe 3d (triangles) electronic distribution curves in OB of the features towards low binding energies from the
crystalline Al7Cu2Fe. The full vertical line at the origin
represents the Fermi level, which is located within 0.1 eV for
maximum in the experimental curves of panels a and b
Al and 0.3 eV for Cu and Fe respectively are satellite contributions to the main peak.
The situation is less simple when the sample contains
elements with open shells. The initial state of XES is
then the same as the final state in PES. Therefore many
features are observed on the emission spectral dis-
tribution curves, which arise from final state effects due
to both ionized and excited configurations. Figure 13
displays the Ce 4f--3d emission spectrum of CeNi2 .
Structures due to transitions involving excited and
ionized 3d 9 4f 1 , 3d 10 4f 0 , 3d 9 4f 2 configurations can be
recognized. In this example, the occurrence in the
spectrum of peaks and bumps (arising from peaks of
smaller intensity) characteristic of both Ce3þ and
Ce4þ established that the CeNi2 intermetallic is
actually a mixed valence alloy (Sonder, 1990).
Figure 18 OB and UB contributions in an icosahedral Al-Pd-Mn quasicrystal as obtained from XES and XAS experiments,
respectively. Small dots: Al 3s-d, full line: Al 3p, diamonds: Pd 4d-5s, triangles: Mn 3d-4s, large dots: Pd d-s, open dots: Mn d-s,
broken line: Al p. All the curves are normalized to their maximum intensity except for the Al p distribution which, assuming the
transition probabilities are constant, is adjusted to be the same intensity at EF as its OB counterpart
rather extended UB Al p counterpart above EF is 19). Note that as in XAS, there is a remaining core hole
observed. This is in line with the high resistivity of this in the final state, therefore this spectroscopy is also site
intermetallic alloy (about 1000
cm at room tem- selective.
perature at which the spectroscopic measurements are The energy loss is measured with respect to the
made) (Belin et al., 1994; Belin-Ferré and Dubois, elastic peak in the reflection or the transmission modes
1996). for small scattering angles. The measurements give the
Note that XAS which refers to an energy range of a fraction of the incident electron beam with energy E
few tens of eV above the absorption edges is often which is scattered into a solid angle d
with the energy
labelled as XANES, namely X-ray absorption near- from E1 to E1 þ. If we denote by q the momentum
edge structure in contrast to EXAFS, extended X-ray transferred from the incident to the scattered beam,
absorption fine structures, which is concerned with then the cross-section of EELS is given by:
energy ranges of a few hundreds of eV above the edges
and mainly carries structural information. d2 s=d
dE / ð1=q2 Þ Imð1=EÞ ð11Þ
3.2 Electron Energy Loss Spectroscopy Imð1=EÞ is the so-called energy loss function and E is
A mono-energetic incident electron beam directed into the dielectric constant which is connected to the
a solid is slowed down because of collisions, either photoabsorption coefficient: For energies around
elastic or inelastic, with the core electrons. When the 10 eV, the loss function is mainly determined by the
incident energy is high enough, ionizations and collective excitations of the valence electrons, whereas
excitations of the electrons of the solid can be for higher energies core excitations are taken into
produced, and inner shell electrons can be promoted account. Finally, in the independent particle approx-
to unoccupied states of appropriate symmetry. Thus, imation, and for small momentum transfers,
electron energy loss spectroscopy (EELS) allows one to
explore the distribution of unoccupied states (Figure d2 s=d
dE / i,f jh i jqrj f ij2 dðEf Ei --
hoÞ ð12Þ
148 Bonding and Stability