HAL Id: jpa-00210537
https://hal.archives-ouvertes.fr/jpa-00210537
Submitted on 1 Jan 1987
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
Theory of f-electron photoemission in light rare-earth insulating systems
J.C. Parlebas, T. Nakano, A. Kotani
To cite this version:
J.C. Parlebas, T. Nakano, A. Kotani. Theory of f-electron photoemission in light rare-earth insulating
systems. Journal de Physique, 1987, 48 (7), pp.1141-1146. �10.1051/jphys:019870048070114100�. �jpa-
00210537�
Theory of f-electron photoemission
in light rare-earth insulating systems
J. C. Parlebas (*), T. Nakano (+ ) and A. Kotani (+ )
L.M.S.E.S. (UA CNRS N° 306) Université Louis Pasteur, 4, rue Blaise Pascal, 67070 Strasbourg Cedex, France
(+ ) Department of Physics, Faculty of Science, Osaka University, Toyonaka 560, Japan (Reçu le 27 octobre 1986, accept6 sous forme définitive le 9 mars 1987)
Résumé.
2014Nous présentons une méthode pour calculer les spectres de photoémission directe f dans le cas des
systèmes isolants de Terres Rares légères ; pour ce faire nous utilisons le modèle d’impureté d’Anderson avec une bande pleine et un état de base qui est un doublet de spin. Nous explorons l’allure générale des spectres en fonction des divers paramètres du modèle. Nous proposons d’expliquer ces spectres à partir de la limite Uff infinie qui est entièrement analytique. Un exemple typique de comparaison de nos résultats avec les spectres expérimentaux de photoémission directe est le cas du composé Ce2O3.
Abstract.
-We present a method for calculating f-electron photoemission spectra of light rare-earth insulating systems within the filled band Anderson impurity model and starting from a spin doublet ground
state. We explore the overall lineshapes of the spectra with respect to the various parameters entering the
model. An analytical explanation of the calculated spectra is proposed in the infinitely large Uff limit.
Typically, contact can be made with experimental direct photoemission spectra in the case of Ce2O3 compounds, for example.
Classification
Physics Abstracts
79.60E
1. Introduction.
The splitting of 4 f-derived photoemission spectra due to the mixing of the 4 f configurations in the ground state (i.e. due to the hybridization in the
initial state which induces the mixed valence effect)
has been well obtained in some compounds of heavy
Rare Earth elements like Sm, Eu, Tm and Yb [1].
For Sm and Tm, the valence deduced from the
experimental photoemission (XPS) spectra is in agreement with the valence deduced from the macro-
scopic measurements (susceptibility, lattice par-
ameter...). For light rare earth systems, especially
for Ce metal and metallic compounds, the situation
is more complicated, first of all, because it is then hard to separate the f contribution from the other conduction states contribution in the XPS spec-
trum [1] unless resonant XPS is used. Nevertheless,
direct f photoemission has often been used to study
the position and occupancy of the f level in Ce
(*) Present address : c/o Prof. P. Fulde, Max-Planck-
Institut, Heisenbergstrasse 1, Postfach 80 06 65, 7000 Stuttgart 80, F.R.G.
-A von Humboldt fellow.
metallic compounds [2, 3]. This kind of experiments
has been rather well understood theoretically within
the impurity model in the limit of large degeneracy (Nf - oo ) [4].
In this paper we focus attention on the f-XPS spectra in light rare-earth insulating systems, es- pecially in Ce203 compounds. One of our motiva-
tions was the recent experimental XPS spectra for a Ce02 film converted progressively to Ce203 [5]. In
the case of Ce02, a cluster model [6] has first been proposed to calculate core level spectra, then the impurity model has been used by several authors [7- 10] ; in particular Wuilloud et al. [7] calculated and discussed the f-spectra of Ce02 compounds but the
case of Ce203 has never been investigated theoretically. In section 2 of the present paper, using
the filled band impurity model we develop an exact
formulation of the f-photoemission spectrum in a
Ce203 type compound with a spin doublet ground
state. Then (Sect. 3) we present an analytical calcu-
lation of the f-XPS spectrum in the limiting case of large electron-electron interactions (Uff -+ oo). In
section 4 we present our main numerical results on f-
Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:019870048070114100
1142
XPS spectra and section 5 is devoted to some
discussion and concluding remarks.
2. Formulation.
We consider a system consisting of a filled valence
band (VB) and 4 f levels of energy sf with spin (but
without orbital) degeneracy and located on the photoexcited site. During the f-photoemission pro-
cess, the Rare-Earth compound will be described by
the following general Hamiltonian (valid for both
initial and final states) :
where
Here atu and aku are, respectively, the creation and annihilation operators of the VB electron with energy Ek (k =1, ... N ) and spin a ; afQ and
af. are those of the 4 f electron with energy Ef and spin a (i.e. Nf = 2) ; Uff is the Coulomb interaction between 4 f electrons and V kf is the hybridization between the 4 f and VB states.
For simplicity we disregard the f orbital degener-
acy in equations (2.1) and (2.2) because it is not so
essential, in a first step, for insulating systems in
contrast to metallic systems (see Ref. [10] as well as
the present Sect. 3). Considering a system state with
one electron in the 4 f level (4 fl configuration), the physical meaning of the energy E f appears in the difference 6f - Ek which expresses the total energy
required for the transition 4 f - 4 f2 corresponding
to the excitation of a valence electron of energy E k to the 4 f level and in the special case of V kf
=0 and Uff
=0. In order to diagonalize H, it is convenient to rewrite H in the form :
where Ho is the same as Ho except that sf in (2.2) is replaced by s f + Uff. We can easily diagonalize
Ho as follows :
by the transformation :
where En is given by the solution of
and
In the case of a filled VB it is very convenient to introduce a
«Slater determinant » state 0 which is defined by all VB and f states being occupied :
and the corresponding energy E, is given by :
Equation (2.10) means that in the special case of our
filled system there is no energy variation which would result from the presence of hybridization.
In the initial state of the photoemission we suppose there is essentially one hole, Ef’ Thus, the ground
state is a spin doublet state deduced from the previously considered I t/1 > state by :
and the corresponding ground state energy is :
where £g is here the highest energy level among
{£n} of equation (2.6) and which automatically
includes Uff. By absorbing a photon of energy v, an f electron is emitted with energy E. So in the final state the two-hole ( I m) ) eigenstates of H are given by :
From Him>
=Em 1m> , Em and B£, are determined by the following equations :
where :
Neglecting here emission from the valence band (see
Ref. [12]), the Hamiltonian describing the direct
transition from the initial state to the final state of
the f-electron photoemission process is the follow-
ing :
where MD denotes the direct f transition matrix element. For singlet final states {I m) }, the f-
spectrum is written as :
where w is the binding energy of the photoemitted
electron :
and r represents the spectral broadening due to the
finite lifetime of the f-hole, as well as to the experimental resolution width. The matrix element
Before testing equation (2.18) numerically let us
calculate the f-XPS spectrum in the analytical limit of Uff -+> 00.
3. Analytical limit.
/
In the case of Uff -+> oo it is convenient to write the
ground state as :
where the state 1°) of energy Eo is defined by all VB
states being occupied. The ground state energy of
1 .0 0) is then just given by Eo + Ef. Furthermore the f-XPS spectrum is expressed as :
with z being here the opposite of the binding energy defined in equation (2.19) :
Taking account of our simplified ground state, equation (3.2) can be rewritten immediately as :
Because of Uff -+ 00, hybridization V kf of equation (2.2) cannot play any role in the initial state of the photoemission process. However V kf can
again act in the final state of the photoemission
process when the f-electron has been photoemitted
and this final state effect can be exactly represented by the self-energy part -V(z) :
with
Equation (3.6) can also straightforwardly be written
as :
moreover, if we assume a constant hybridization, Vkf
=V / BIN (cf. [6] and [71), X(z) is essentially given by the local Green function associated with the valence band :
In equations (3.7) and (3.8) the factor of 2 is in fact the degeneracy N f. It is interesting to notice that the
degeneracy effect only appears associated with
V 2, at least in the present analytical limit (Uff -+ oo ),
so that the spectrum is not changed with N f, provided that N V2iS fixed. For finite Uff it has been
shown (for core photoemission) that the essential features of the spectrum are not much changed with N f if Nf V2 is fixed, although some minor changes happen to show up [10]. Our f-XPS spectrum (3.5) is formally given by the same expression as that used
for the f-local density of states in (an insulator
version of) the Anderson f impurity
-or f extra
orbital-model (see for example [11]) : there might be
two bound states (b.s.), one b.s. or no b.s. at all depending on the model (V, Ef, centre, shape and
width of the VB) and according to the following equation
where GR is the principal part of G.
4. Numerical analysis and discussion.
In this short paper we would like to analyse our
numerical results for direct f-photoemission with respect to the main parameters which enter our model. Complementary results as well as a detailed
fit to available experimental data will be presented
in a further publication [12]. In our calculation, we
treat the filled VB as a finite system consisting of N
discrete levels with equal energy spacing
1144
where W is the VB width and we put W
=2 eV except when we study the VB width on the spectrum.
The cluster model and the band model correspond to
the limits of small and large N, respectively. The hybridization Vkf is assumed to be Vkf
=V/ J N
with a constant V as already assumed in section 3.
In figure 1 we show the dependence of the calcu-
lated photoemission spectra F (w ) on the number N, by using V =1 eV ; Uff = 6 eV ; F = 0.5 eV and Ef
=0.5 eV where E f and also the binding energy w
are measured with the centre of the VB as origin.
First of all we recover that F (w ) depends on N for
N , 3 but almost converges for N ;:= 4 [7, 8]. In the following calculations we choose N = 4 and F
=0.5 eV.
In figure 2, we show the behaviour of photoemis-
sion spectra F (to) for various values of the hy-
bridization V. According to the
«extraorbital
»pic-
ture (see Sect. 3), in order to reproduce a two-peak
structure, V must be sufficiently large (V
>0.5 eV :
see Fig. 2b). Actually in our rectangular (discrete)
band model we always get two b.s. on each side of the VB but in order to obtain sizeable weights of
these b.s., V has to be sufficiently large. For small
values of V (Fig. 2a), the spectra essentially show one-peak structures corresponding to a resonance
within the VB and near Ef. The role of r is to broaden the structures especially in the case of b.s.
outside the VB.
Fig. 1.
-f-photoemission intensity versus binding
energy w for various values of the number N of the electronic levels in VB. Other parameters are fixed as
ef = 0.5 eV, W = 2 eV, V =1 eV, Uft = 6 eV; Ef and w
are measured with the centre of VB as origin ; a Lorentzian broadening r
=0.5 eV has been used (see Eq. (2.18)).
Fig. 2.
-f-photoemission intensity versus binding
energy w for various values of the hybridization V = 0.1, 0.3, 0.5 (eV) (Fig. 2a) and V = 0.6, 0.8, 1.0 (eV) (Fig. 2b). Other parameters are fixed as sf
=0.5 eV ;
W = 2 eV and Ua = 6 eV.
In figure 3 we test the sensitivity of the. two-peak
structure upon the position E f of the f level for a given large hybridization (V = 1 eV). Our results
can be understood by the analytic expression (3.5) of F (Cù ) in the limit of U ff -+ oo. By defining -V (w ) = + i y (co ), we obtain the relations 5 (to ) =
-