HAL Id: jpa-00246769
https://hal.archives-ouvertes.fr/jpa-00246769
Submitted on 1 Jan 1993
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.
Jellium sphere model in overlayer-substrate systems
Chen Lu-Jun, Wang Ning, Luo Enze
To cite this version:
Chen Lu-Jun, Wang Ning, Luo Enze. Jellium sphere model in overlayer-substrate systems. Journal
de Physique I, EDP Sciences, 1993, 3 (4), pp.1053-1058. �10.1051/jp1:1993176�. �jpa-00246769�
Classification
Physics
Abstracts73.20Hb 73.30Y 79.60Gs
Jellium sphere model in overlayer-substrate systems
Chen Lu-Jun
(I), Wang Ning (2)
and Luo Enze(3)
(1)
Department
ofPhysics,
Box 271, XidianUniversity,
Xi'an, 710071, China (2)Beijing
Vacuum Electronic Devices Research Institute,Beijing,
100016, China (3)Department
ofPhysics,
Box 271, Xidian University, Xi'an, 710071, China(Received 22 May 1992, revised 30 September 1992, accepted 16 December 1992)
Abstract. It was found in calculations that the
jellium/slab
model reduces theability
of the adsorbed atoms to attract the valence electrons. As a result, the calculated minimum of the work function is too lowcompared
witli the experiment. In this paper, as animprovement,
weproposed
a
jellium sphere/slab
model, in which thepositive charge background
is described by discretejellium
sphere. Test calculations show that this treatment lifts the work function minimum by 0.4eV and 0.8eV for Cs/~V(001) and Cs/Ir(001) system,respectively, resulting
in better agreement with experiments.1. Introduction.
Using
thejellium/slab
model[II
and the Film-LAPW method[2-5],
the coveragedependence
of work
functions,
energy bands and surface states for alkalimetal/transition
metalsystems
can be well described
[6-8]. However,
the effect of discretecharge
distribution of adsorbed atoms in those systems on the work function is very remarkable. As a matter offact, Lang [9]
used the
pseudopotential
in hisjellium
model to simulate the discreteness for a bare metal surface. The first orderpseudopotential
corrections (&W)
of work functions are different for variouscrystal
faces. Forexample,
the corrections for bare surfaces ofCs(110), Cs(100)
andVs(I it)
are0.23,
0.61 and 0.67 eVrespectively.
Thenegative correction,
in the baresurface case, shows the decrease of the
dipole
moment which is formedby
thepositive
background
and the electronsspreading
out to the vacuumregion.
This shows also that theattracting ability
of ion-latticerepresented by
thepseudopotential
isstronger
than that of uniformbackground. Moreover,
we consider that as the discreteness in both substrate andoverlayer
areneglected simultaneously
inLang's jellium/jellium
model[10],
the result in thiscase is consistent with the actual case in the sense that
positive
discretecharges
in bothregions
have an
opposite
andnearly
counteractive force on electrons.Therefore,
the better calculatedresults,
were obtained in that time.However,
this counteraction does not yet exist theieilium/slab model,
which leads to a remarkable influence on the calculatedquantities,
such as1054 JOURNAL DE
PHYSIQUE
I N° 4the work function etc, because of
unilaterally ignoring
the discreteness ofpositive charge
in theadlayer.
Hence, thejellium sphere/slab
model isproposed
in this paper to simulate the discreteness ofpositive charge
inoverlayer.
Thecorresponding
mathematic treatment for thismodel is
given
below. Calculation results show that this model has wellimproved
thecalculated work
function,
which is the most sensitivequantity
among all calculatedquantities.
2. Jellium
sphere
model and it's mathenlatic treatment.In order to consider the discreteness of
adatoms,
but withoutexpanding
the calculation unit because of thecomputational capability,
the totalcharge
in thejellium
of a cell is concentrated in ajellium sphere
which is laid on a 2-dimensional cell as shown infigure la,
and we assumethat the
charge
ishomogeneously
distributed in thesphere.
The main purpose of this paper is to get a clearunderstanding
of the effect of discretecharge
on the work function and to knowwhat an
important
role itplays
in theadsorption
process.The parameters are calculated
according
tofigure16. Dj
isgiven
to be the thickness ofjellium
and R to be the radius ofjellium sphere
which ispositioned
at z = zo. We have zo = zi +Dj/2,
where z= zi denotes the interface between the interstitial
(TNT) region
and thejellium-vacuum region. r(z)
infigure16
isgiven by
r(z)= ~/R~- (z-zo)~ zo-R«zwzo+R.
z
z
~ ~
iy
3j
j i
ii
~R
t~iJflilt~
ai bj
Fig.
I. al The schematic drawing ofjellium
sphere/slab model b) The schematic drawing for parameters calculation.Suppose
Qvai is the number of valence electrons and is filled in thejellium sphere.
Then at the coverage @, the number of electrons in thesphere
is Q~~i.They
distributehomogeneously
in thejellium sphere
and thusPi =
~~~ (in sphere)
p =
wR3
3
P2 "
°
(out
ofsphere)
which can be
expanded
further into the 2-dimensional star functionsP
(rj
=jj pG(zj
q~G(rjjj
GE jG,j
with
N~
PG(Z )
"
$
P (~) Pi
(~(( d~((s
V~G(ri =
~~
~j~
e'~~
~< <R>In our
problem, t~
=
0 and G
=
(R t~)
is the discrete part of the 2-dimensional space group,Nop
is the order of the group. The substitution ofq7~(r)
intop~(z) gives
PG(Z)
"(~
i jj~
PI ~ '~~ '~ ~~(ReG
~~~~RG)~,
G~
=(RGIy
~~ld~~~~~
~~~~~~~°~~~~~~
~~~~~~~~~~~z j
=
ii ~j
~j~
i
SiniGx /
~~~~~~
'
~'~~~~
~
~~~~~~~
when
G~
= 0 orG~
=0,
the result ofp~(z)
isjust
the same as the limitation of the aboveequation
atG~»0
orG~»0,
I-e- when G=
0, p~(z)
isequal
to~°~~ ~~~~~°~~ ~j~ "~
~~
~i~~~
~
=
~~~~
"[R~
(z zo)~]
4
~~3
A3
This distribution is a
parabolical
curve, I-e- the first term inexpansion
of p(r corresponds
to aninhomogeneous jellium
flake with aparabolical
distribution.3. Results and
analysis.
In order to
investigate
how remarkable the influence of the discreteness ofpositive charge
in theadlayer
on the calculatedquantities,
such as the work function etc.,is,
and what animportant
role the discretenessplays,
theCs/Mf(001)
system andCs/Ir(001)
system arecalculated under some coverages
respectively.
Thecrystal
lattice constants for W and Ir are chosen to be 5.9813 a-u- and 5.1212 a-u-respectively.
The valence electrons for both of themare taken to be
5d~6s~
and5p~5d~6s~,
and core electrons to be[(Xel4fl~]
and[(Xe-sp~l4f~~]
respectively.
Themuffin-tin(MT)
radii are 2.55 a-u- and 2.50 a-u-, and the radii ofjellium spheres
are 2.5 a-u- and 2.4 a-u-respectively
for both systems. The calculation results aregiven
infigure
2 and table I. It can be seen that the discreteness has resulted in about 0.4 eV,arising
from the work function minimum for theCs/Vf(001)
system(at
m
0.55,
orNaflQsm0.21),
and about 0.8eVarising
for theCs/Ir(001)
system(@=0.6,
orNa/Ns
»
0.17).
Inpractice,
the calculated work functions forCs0V(00 Ii
andCs/Ir(001) using
thejellium/slab
model are 1.34 eV and 0.2 eV at= 0.5 and
= 0.6
respectively.
When1056 JOURNAL DE PHYSIQUE I N° 4
o
(cv)
2
0.1 0.2 0.3 0.4
Na/Ns
Fig. 2. The
comparison
of calculated results I) Calculation curve of CsliV (001) system by flake model in reference [8] and (O) is a calculated value at e=
0.5 by
sphere
model ;2)Experimental
curve of Cs/~V (001) systemreported
in reference [I Ii 3) Calculation curve of Csflr (001) systemby
flake model in reference [7] 4) Calculation points(symboled
by (o)) of Csflr (001) system, the upper branchby
sphere model and the lowerby
flake model, (li) denotes theexperiment
value (Ref. [6]).Table I. The
comparison ofcalculated
resultsby
thejellium/slab
and thejellium sphere/slab model,
where Na/Ns = 0.283for Cs/Ir(00i)
andNa/Ns
= 0.387
for Cs0V(00i).
~ Jellium
sphere
Jellium flaked System ~~
fi~~ ~fi
(eV) Q~~
W(eV) Q~~
~VUC0. ii 0.4 Ir-Cs I.4i 0.460 0.98 0.44i 4.3 §b
0.17 0.6 Ir-Cs 1.02 0.515 0.20 0.470 9.7 §b
0.19 0.5 W-Cs 1.74 0.740 1.34 0.685 2.7 §b
using
thejellium sphere/slab model, they
become 1.74eV and 1.02 eV. Theexperimental
work function minimum is 1.3 eV forCs/Ir(001) [6].
This indicates that the contribution ofdiscrete
charge
distribution inadlayer
is veryimportant
for work function determinations.The calculated values
by
thejellium sphere/slab
model must have asensitivity
to thechange
ofjellium sphere
radiusR, since,
forexample,
the work function minimum ofCsflr(001)
should bechanged
to 0.2 eV from 1.02 eV when thejellium
is transferred to the flake from thesphere.
However,
the work function is insensitive to thechange
of the radius in thelarge
valueregion,
at least, say, in the
region
from R= 1.5 a.u, to R
= 2.4 a.u. When the radius R is reduced to
1.5 a,u., the calculated work function and the total
charge
in vacuum at=
0,4 are 1.42 eV and 0.453 e
respectively
which are much closer to the valuesII.41eV
and 0.460e)
obtained1/2
by
R= 2.4 a,u. Since
self-consistency
wasassumed,
when AV ~ dz ~ 0.03R~d,
inour
calculations,
anuncertainty
of about ± 0.2 eV for the work function may existaccording
toour
experience.
That is to say, a variation of thejellium sphere
radius within a range from 1.5 a-u- to 2,4 a,u, does notobviously
influence our results.Some
change
of valence electroncharge
in thejellium-vacuum region
has arisen from thediscreteness,
and thecharge
redistribution has resulted(see Fig.
3 andFig. 4).
Thecharge
redistribution can be considered as the result of the fact that the
attracting ability
of thejellium sphere
is stronger than that of the uniform flake. Thenegative
centre of electrons in theregion
between
adlayer
and substrate surfacelayer is,
in thesphere model,
closer to the centre ofpositive charge
inadlayer
than that in the uniform model.According
tofigure
3 andfigure 4,
itcan be considered that there exist three
dipoles (as schematically
shown inFig. la)
in the surfacelayers,
which influence calculation work functions. The first(pi )
and the second(p~ point respectively
to the slab and thejellium sphere
from thenegative
centre of electrons in theregion
between theoutlayer
slab and thejellium sphere.
The third(p~) points
to thejellium sphere
from the centre of electrons in the outer half-side of thejellium layer
and thosespreading
into the vacuumregion.
The contributions of the threedipoles
to the work functionare
symbolized
as ~fii, W~ and W~
respectively, then,
the total contribution can beregarded
asAW
= W
i + W~ ~fi~. When the
negative
centre isgetting
closer to thejellium sphere (I.e.
far away from theslab), Wi
will beincreasing
and W~decreasing.
While W~ is notremarkably changed according
to thecharge
redistribution in the outer half-sidejellium layer
and in thevacuum
region
infigure
3. Hence, the total contribution leads to a net increase of the workfunction. In other
words,
the attractiondiscrepancy
between thejellium sphere
and flake to the electrons has taken on animportant
role in the work function. Less than lo §b of thechange
of electroncharge
has made the work function extend to alarge region (see
Tab.Ii.
Indeed aquite precise description
of thepositive charge
distribution in thejellium-vacuum region
is very necessary. Of course, there are other factors which could influence the calculated work function[8].
The mainsignificance
of this paper is make usknow,
that the discretecharge distribution,
beside the other factors is animportant
factor in the work function. Once the other factors have been taken into account, there is a veryimportant
reason forconsidering
the effect ofpositive charge
distributionby
thejellium sphere/slab
model. Here we stressagain
theimportant
effect of discreteness and thecorresponding
treatment.In
Summary,
as thejellium/slab
model reduces theability
of the adsorbed atoms to attract the valenceelectrons,
thejellium sphere/slab
model isproposed
to simulate thepositive
discrete
charge
distribution of the true adatoms in theadlayer.
The calculation shows that thelip(Z)
x 10~a,u.~~
l,
p(0.5)-p(0.0)
flake model,
2
1',
2,p(0.5)-p(0.0)sphereinodel
D,
o 50 loo x o.15 a-u- Z
Fig.
3. Thediscrepance
curves of valencecharge
distribution in thejellium-vacuum region
(I.e. in theadlayer) by sphere
and flake models for the Cs/~V(001)
system at= 0.5.
JOURNAL DE PHYS>QUE -T 3, N'4, APR>L >993
1058 JOURNAL DE
PHYSIQUE
I N° 46p(Z)
x 10~a.u.~~
lo
o
Fig.
4. The contour line ofcharge density discrepany
Ap(= p~~jj,~~~~~ p~~j~~~w~) on the
profile
W (I lo) in a had calculated cell for CsliV
(001)
system at coverage e= 0.5.
influence of discreteness upon the work function is very
important
andgives
a contribution ofabout 0.4eV~0.8eV to the work function minimum for different
overlayer-substrate
systems.
This shows us thatcharge
discreteness inadlayer
is animportant
factor ininfluencing
the determinations of work function.
References
[Ii WANG N., CHEN K. L., WANG D. S.,
Phys.
Rev. Lett. 56 (1986) 2759.[2] KRAKAUER H., POSTERNAK M., FREEMAN A. J.,
Phys.
Rev. B19 (1979) 1706.[3] POSTERNAK M., KRAKAUER H., FREEMAN A. J., KOELLING D. D.,
Phys.
Rev. B 21 (1980) 5601.[4] OHNISHI S., FREEMAN A. J., WIMMER E.,
Phys.
Rev. B 29 (1984) 5267.[5] WIMMER E., FREEMAN A. J., HISKES J. R. et al.,
Phys.
Rev. B 28 (1983) 3074.[6] Wu Ru-QIAN, WANG D. S.,
Phys.
Rev. B41(1990) 12541.[7] Wu Ru-QIAN, CHEN K. L., WANG D. S., WANG N., Phys. Rev. B 38 (1988) 3180.
[8] CHEN Lu-JUN, WANG NING, WANG D. S., ENzE LUO, Phys. Rev. B 44 (1991) 8942.
[9] LANG N. D., KOHN W.,
Phys.
Rev. B 3 (1971) 1215.[10] LANG N. D., Phys. Rev. B 4 (1971) 4234.
jl Ii SWANSON L. W. and STRAYER R. W., J. Chem.