Photoelectron spectroscopy study of the 5 f localization in Pu
L. Havela,
1,2
T. Gouder,
1,*
F. Wastin,
1
and J. Rebizant
1
1
European Commission, Joint Research Centre, Institute for Transuranium Elements, Post Box 2340, D-76125 Karlsruhe, Germany
2
Department of Electronic Structures, Charles University, Ke Karlovu 5, CZ-121 16 Prague 2, Czech Republic
Received 7 January 2002; published 12 June 2002
X-ray photoelectron spectroscopy and high-resolution ultraviolet photoelectron spectroscopy studies were
performed on thin Pu layers and compared with the studies on bulk surfaces prepared at various temperatures.
We followed the process of reconstruction of the -Pu surface into a -Pu type. Its kinetics depends strongly
on temperature. The electronic structure of Pu layers varies markedly with the thickness and the localization of
the 5 f states can be achieved in the case of ultrathin Pu layers deposited on Mg. The valence band displays for
the majority of Pu systems three sharp features within the first eV below the Fermi energy, the position of
which is fixed. The localized 5 f states yield a broad maximum at 1.6-eV binding energy.
DOI: 10.1103/PhysRevB.65.235118 PACS numbers: 71.20.Gj, 71.28.+d, 79.60.-i
I. INTRODUCTION
5 f electronic states in light actinides Th-Pu have a band
itinerant character and contribute to metallic bonding, in
analogy to the d states in transition metals. Starting with
americium ( Z =95), the heavier actinides adopt a localized,
atomic character, analogous to the situation of the 4 f states
in lanhanides, and thus the 5 f electrons do not contribute to
cohesion.
It has been generally accepted that high-resolution
valence-band photoelectron spectra of actinide metals exhibit
a gradually increasing 5 f spectral intensity in the sequence
Th, U, Np, Pu.
1
It is broad, has no distinct sharp features,
reaches maximum at the Fermi energy E
F
, and indicates the
itinerant character of the 5 f states. The 5 f localization in Am
leads to a shift of the 5 f emission from the Fermi level and
the appearance of a multiplet structure. Thus photoelectron
spectroscopy proved to be a very convenient tool to distin-
guish between the band and localized situations.
Being right at the localization threshold, Pu has an espe-
cially interesting position within the actinide series. From the
point of view of crystal structure and metallurgy, the com-
plexity of Pu has no parallel among elements of the periodic
table. It exhibits six allotropic modifications and some of the
structures have a very low symmetry.
2
Remarkable volume
variations between the phases must be due to the varying
balance between 5 f bonding and localization. The loss of the
5 f bonding means a loss of a part of the cohesive energy and
leads to the higher atomic volume. At the same time, the
system gains energy coming from Coulomb and spin-orbit
interactions by more correlated motion of electrons in
atomiclike 5 f states. The low-temperature phase -Pu
monoclinic, stable up to 395 K, can be described as a
5 f -band system, albeit with strong electron-electron correla-
tions. The other most prominent phase is -Pu, stable be-
tween 592 and 724 K, which has the largest volume by 20%
larger than the phase. As it can be stabilized in a larger
temperature range by doping, low-temperature data are also
available.
3
The volume variations are strongly affecting the
overlap of the 5 f wave functions centered on neighboring Pu
ions; the larger overlap increases the width of the 5 f band,
or, more generally, tends to a more delocalized character of
the 5 f states. The weakly magnetic character of - and -Pu
are similar, but local-density approximation LDA calcula-
tions of -Pu assuming 5 f itinerancy have difficulty in ac-
counting for the cohesion properties.
4–7
New theoretical ap-
proaches, emerging recently to account for -Pu properties,
assume that part of the 5 f states are localized, whereas the
itinerancy is still preserved for the remaining states,
4
or span
the Mott transition using a dynamical mean-field theory.
7,8
The rather simple picture of featureless spectra of the 5 f
band mentioned above was obtained for the -Pu phase.
1
The valence-band spectra of the phase exhibit a Fermi-
level peak narrower than for -Pu, and there is another dis-
tinct maximum at 0.8-eV binding energy.
9
The dramatic dif-
ference between the valence-band spectra of - and -Pu was
not confirmed by later experiments.
10
In this latter work the
-Pu features shown in Ref. 9 are reproduced, but -Pu is
strikingly similar, also displaying another maximum at 0.8
eV, which is only somewhat weaker than for -Pu. In recent
paper
11
we indicated that a surface reconstruction of -Pu
can be responsible for the disagreement. Moreover, we
showed that the electronic structure of Pu thin layers varies
dramatically when reducing the thickness.
In this paper we describe photoemission studies on bulk
-Pu with the surface prepared under various conditions,
which provides information about where differences in -Pu
spectra may originate. We demonstrate how the increasing
degree of 5 f localization, which is the main reason for dif-
ferences between the - and -Pu, can be extended for ultra-
thin Pu films, reaching the limit for one monolayer. A
complementary information about the 5 f delocalization is
provided by the Pu-4 f core-level spectroscopy. The final part
discusses the results obtained in the context of other Pu
based systems. We could identify three types of features,
belonging to the localized 5 f states, to band 5 f states, and
general features appearing at an intermediate delocalization.
II. EXPERIMENT
One of major difficulties of the work with Pu is the ra-
diotoxicity, which precludes work with an open material in
laboratory conditions, and which is restricting the amount of
material handled even in a dedicated laboratory. The work
described here was performed in a photoelectron spectrom-
eter equipped with Leybold LHS 10 hemispherical analyzer,
PHYSICAL REVIEW B, VOLUME 65, 235118
0163-1829/2002/6523/2351189/$20.00 ©2002 The American Physical Society 65 235118-1