HAL Id: jpa-00208545
https://hal.archives-ouvertes.fr/jpa-00208545
Submitted on 1 Jan 1976
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, estdestinée 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.
Neutron diffraction studies of liquid silver and liquid Ag-Ge alloys
M. C. Bellissent, P. Desre, R. Bellissent, G. Tourand
To cite this version:
M. C. Bellissent, P. Desre, R. Bellissent, G. Tourand. Neutron diffraction studies of liq- uid silver and liquid Ag-Ge alloys. Journal de Physique, 1976, 37 (12), pp.1437-1444.
�10.1051/jphys:0197600370120143700�. �jpa-00208545�
NEUTRON DIFFRACTION STUDIES OF LIQUID SILVER
AND LIQUID Ag-Ge ALLOYS
M. C. BELLISSENT
(*)
and P. DESRELaboratoire de
Thermodynamique
etPhysico-Chimie Metallurgiques
associé au CNRS
(LA 29)
ENSEEG Domaine Universitaire BP44,
38401St-Martin-d’Hères,
FranceR. BELLISSENT and G. TOURAND
Service de
Physique
du Solide et de RésonanceMagnétique,
Centre d’Etudes Nucléaires de
Saclay,
BP n°2, 91190 Gif-sur-Yvette,
France(Reçu
le3 juin 1976, accepté
le16 juillet 1976)
Résumé. 2014 Une étude structurale de l’argent liquide à 1 323 K et des alliages Ag-Ge à 1 123 K a été réalisée par diffraction de neutrons. On a obtenu par transformation de Fourier les fonctions de distribution de paire g(R) et calculé les nombres de
coordinance.
Les fonctions d’interférence expé-rimentales sont ensuite comparées à celles déduites d’un calcul du type Percus-Yevick. Enfin les fonctions d’interférence partielles ont été évaluées, dans l’hypothèse de leur invariance avec la
concentration, pour les paires
Ag-Ag
et Ge-Ge. On montre qu’elles sont en très bon accord avec les facteurs de structurerespectifs
de l’argent et du germanium pris isolément à l’état liquide.Abstract.
2014 Structural studies of liquid Ag at 1 323 K and liquid Ag-Gealloys
at 1 123 K have beencarried out by means of neutron diffraction. The pair correlation functions were evaluated by means
of a Fourier transform and coordination numbers were calculated. The interference functions are
compared with those deduced from Percus-Yevick calculation. Using the
assumption
of concentra-tion independency, partial interference functions are evaluated for Ag-Ag and Ge-Ge
pairs.
Theyare shown to be very similar to the
respective
structure factors of liquid silver andliquid
germanium.Classification
Physics Abstracts
7.126
1. Introduction. - There have been
relatively
fewstudies of
liquid binary alloys by X-ray
or neutrondiffraction. The
experimental
difficulties encountered whenworking
with suchalloys (chemical reactivity
at
high temperature,
use ofisotopes, etc.) require
structure studies to be carried out on a
typical
metallicsystem
representing
a class whosemacroscopic
pro-perties
have certain common characteristics.This is the case of
alloys including
a metal of the typegold, silver,
copper, aluminium and a semi- conductor(silicon, germanium)
or a semi-metal(bismuth, antimony).
In fact all these systems have aphase diagram
with asingle
eutectic the fusiontemperature
of which isrelatively
far from that or thepure elements
(important
eutecticeffect).
On the basis of the
thermodynamic properties,
thefusion
heat of the eutectics has been estimated and i is found to beabnormally
low(1).
The structural
study
of condensed matter is mossoften
performed by electron, X-ray
or neutron diffracttion.
Although
the last two types of radiation hav(many common
characteristics, they
differ consderably on
a certain number ofpoints [1].
The presen’study
refers to thesilver-germanium alloy.
Thi:system
was chosen as anexample
among the metal semi conductoralloys
defined above because of it;relatively
favourable chemicalproperties
for studie;at
high temperature.
We also considered that it would beinteresting
tostudy
silver under the samfconditions as this had not been
previously
carried ouby
means of neutronscattering.
The
crystallographic
characteristics of the twc(*) This work has been performed thanks to the financial support of the Pechiney-Ugine-Kuhlmann Society.
(1) Moquet, J. L., Private communication, 1972, LTPCM, Grenoble.
Article published online by EDP Sciences and available at http://dx.doi.org/10.1051/jphys:0197600370120143700
1438
constituents silver and
germanium
arerelatively
different. Silver
crystallizes
in the f.c.c.system (ao
= 4.086 2A)
whilegermanium crystallizes
in thediamond system
(ao
= 5.657 35A).
From the neutron
point
ofview,
silver and ger- manium have coherentscattering
cross-sections ofrespectively
4.6 and 8.8 barns with effectiveabsorption
cross-sections of 36 and 1.3 barns for a
wavelength
of 1.08
A.
On the other hand silver has stableisotopes (117 Ag
and109 Ag)
with coherentscattering
cross-sections of 8.7 ± 0.5 and 2.3 ± 0.2 barns
[2]
suffi-ciently
different topermit
accurate determination of thepartial pair
correlation functionsgpgAg(R ), 9A,Ge(R)
andgGeGe(R)
characteristic of thisalloy.
The
experiments
arecurrently being
undertaken and will bepublished
in the near future.2. Basis of structural studies. - 2 .1 AVERAGE COHE- RENT INTENSITY
(2).
- The waveamplitude
diffractedby
anassembly
of atoms of types a(a
=0, 1, 2, ...)
isexpressed by
theequation :
where
fo, fl,
... are coherentscattering lengths
of the types0, 1,
... and ri(a) represents theposition
of theith atom of
species
a.The average coherent
intensity
isexpressed
asfollows :
If we take rj(fJ) - ri(.) =
Ry
this lastexpression
becomes
since Y y
i sink. Rji
= 0regardless
of the valuesi(a) i(,P)
of
i(,%)
andj(P).
In the evaluation of the
term Y- Y-
cosk . R ji,
twoi(lx) i(P) cases must be considered :
where N is the total number of atoms,
ð(Xp
the Kro-necker delta
symbol and C(X represents
the atomic concentrations suchthat Y
ca =1 ;
(1) While the result of the calculations is well-known, it appears to us opportunate to give the whole development which in our knowledge, cannot be found in the literature.
By analogy
with a system with asingle
consti-tuent
[3],
a group ofpartial
interference functions is defined in accordance with therelationships :
Hence :
r
with
where Q is the total volume of the system, cllj’ the
probability
offinding
an atomfl
at the centre of asmall element of volume dR
knowing
that at the sametime there is an atom a at the
origin, gap(R)
represent-ing
thepair partial
correlation function.Hence we obtain :
The normalization conditions
require
that when R tends towardsinfinity g0152P(R)
becomeequal
tounity
and
80152P(k)
such that :If we assume that the surface
particle gives
anegligible
contribution in the calculation of the
preceding
sum,we obtain
(except
for very small values ofk) :
’The
average coherentintensity
then becomes :2.2 BINARY SYSTEM. - For a
binary
system(a, f3
=1, 2),
the average coherentintensity
is express- ed as follows :Hence the total interference function is deduced
by
the
relationship
or
y
theglobal
interference functionS(k)
and thepartial
interference functions
Siil(k), S22(k)
andS12(k)
relative
respectively
to thepairs 1, 1 ; 2, 2
and1, 2 being
normalized to
unity.
The Fourier transform of the
partial
interference functionsgive pair partial
correlation functionsg(Zp(R)
which
permit
local order to be achieved.We find that to obtain the
partial
interference func- tionsSa.p(k),
we must carry out threeexperiments
withdifferent ca
ffl
coefficients. We canmodify
either theconcentration ca or the nature of the interactions
represented by fo.
The modification of concentrations is the
simplest
method but this
implies the, rarely justified, hypothesis
of the invariance of the
partial
interference functions with respect to the concentration. In any case,initially
and to obtain a first evaluation of the
partial
inter-ference
functions,
thehypothesis
of their invariance with the concentration has been used.In the
present
case of theAg-Ge alloys,
it ispossible
to vary the interaction
intensity
withoutchanging
itsnature. For this it is sufficient to carry out diffraction measurements with different
isotopic
concentrations ofsilver 107 Ag
and’09Ag.
3.
Experimental
method. - 3.1 APPARATUS. - Theexperiments
were carried out as follows :- partly
on the ILL spectrometer D4 on the hotsource
(A
= 0.697A) [4],
- partly
on the spectrometer HIO of the reactor EL3 atSaclay (A
= 1.133A) [5].
The measurements were carried out for wave-
vectors k = 4n sin
OIA, varying
from 0.10A -1
to 14.85A - 1.
The
samples
wereplaced
incylindrical
vanadiumcontainers 7.6 mm in diameter and 0.2 mm thick.
As
regards
thesample
environment conditions :1)
at ILL we used a furnaceconsisting
of a vana-dium
cylinder
25 mm in diameter and 0.1 mmthick, 2)
atSaclay high temperature experiments
werecarried out with an induction
heating device,
thesample being by
itself in the beam.3.2 DETERMINATION OF
IE(2 0).
- In order tomeasure the scattered
intensity IE(2 0)
we used foreach
sample
two methods of dataacquisition depend- ing
on theexperimental
set-up.At ILL three successive spectra were
obtained;
(i) Ii(2 0) intensity
diffusedby
the vanadium fur-nace
alone,
(ii) 12(2 0) intensity
diffusedby
the vanadium container inside the vanadium furnace(3),
and
(iii)
theI3(2 0) intensity
diffusedby
thesample
in the vanadium container inside the vanadium furnace.
At
Saclay only
two spectra were obtained(because only
thesample
is seenby
the neutronbeam) [1];
(i) I2(2 0) intensity
diffusedonly by
the vanadiumcontainer,
and
(ii)
the13(2 0) intensity
diffusedby
thesample
inthe vanadium container.
The determination of this scattered
intensity IE(2 0)
will be carried out in all cases
taking
account of thedifferent
absorption
factors(see appendix 1).
It is then necessary to make a correction within the static
approximation.
This type of Placzek[6]
correc-tion is
developed by
Yarnell[7]
and consists in anevaluation of this
inelasticity
effect. We also estimatedfollowing
Blech and Averbach[8]
the contribution due tomultiple scattering,
but this correction was notapplied,
as its variation of less than1/1 000
for thewhole
angular
range falls within theexperimental
statistical error.
3. 3 DETERMINATION OF INTERFERENCE FUNCTIONS.
- We have determined the total interference func- tions for three different
compositions
of theAg-Ge alloy
and the structure factor of silverS(k) being
normalized
by
thefollowing
relationIf we admit that apart from coherent
scattering
allthe other contributions
(multiple scattering,
incoherentscattering, etc.)
areisotropic
we can write :where X represents the above
isotropic
contribution to the scatteredintensity.
As a result :In
particular :
) (3) In case of heating by Joule effect, using a vanadium resistor,
r the interest of using vanadium is that its coherent scattering cross-
-
section is practically zero. In fact the only contribution is thecompletely isotropic incoherent diffusion, which simplifies the
corrections.
1440
According
tothermodynamics,
for a monoatomicsubstance :
where
kB
is the Boltzmann constant,po is the number of atoms per volume
unit, f3
is the isothermalcompressibility,
and for a
binary alloy :
In this
expression
and
fl
andf2
are the coherentscattering lengths
ofthe constituents 1 and 2.
al is the
activity
of the constituent 1(Its
values taken forAg-Ge alloys
are deduced from acompilation
dueto
Hultgren [9]).
Vm
is the molar volume of thealloy.
The values chosen for
Vm
areexperimental
valuesfor pure
Ag,
0.858 at% Ag
and 0.759 at% Ag alloys.
For the 0.402 at
% Ag alloy,
the value ofVm
resultsfrom an
extrapolation [10].
In addition the value of
1(0)
is determined on the basis of theexperimental
curve which alsopermits
one to obtain the
asymptotic
value of the scatteredintensity I( oo)
since k issufficiently large
for theinterference effects not to be observable.
4.
Experimental
results. - 4.1 LIQUID SILVER. -We have
represented respectively
infigures
1 and 2 thestructure factor and the
pair
correlation functiong(R)
of
liquid
silver at 1 323 K.We compare, in table I
below,
our results with those obtained at the sametemperature by Wagner
et al.
[11]
and at varioustemperatures by
Pfan-nenschmid
[12]
and Waseda et al.[13].
It should be noticed that there exists a
general agreement
with theprevious
results for both theS(k)
and the
g(R)
features.However,
inspite
of thestrong
FIG. 1. - Structure factor S(k) of liquid silver at 1 323 K. Percus-
Yevick calculation (full line). Experimental curve (open circle).
FIG. 2. - Pair correlation function g(R) of silver.
absorption
cross-section ofsilver,
this first structuralinvestigation by
neutron diffractionpresents 4 higher
first
peak
in theS(k)
and ahigher
coordination number : 11.93 which are the main characteristics of acompact
structure in theliquid
state.Moreover,
in the Percus-Yevickapproximation
andusing
a hardsphere potential,
we calculated the struc- ture factor of silver with a hardsphere
diameterof
RAg
= 2.599A
evaluatedaccording
to Aschcroftet al. method’s
[14].
It will be noted that there is a verygood
agreement for both thepositions
and the inten- sities of allpeaks
values and the calculated ones asshown in
figure
1.TABLE I
4.2 SILVER-GERMANIUM ALLOYS. - The interfe-
rence
functions were determined at 1 123 K.They
arerelative to the
following alloys :
a) Ag :
0.858at %
Ge : 0.142at % b)
Eutecticcomposition
Ag :
0.759at %
Ge : 0.241at % c) Ag :
0.402 at%
Ge : 0.598 at%.
Table II summarizes the characteristics of the obtained interference functions which are shown in
figure
3.FIG. 3. - Interference functions of three different Ag-Ge alloys
at 1 123 K. a) Ag,.,,,Geo.112.b) EutecticAgo.7_sgGeo.241- c) Ago.402Geo.598-
As the
germanium
concentrationincreases,
we notethat there is a decrease and a
softening
of the first maximum whichcorresponds
to the transition from ametal
type configuration
to a covalenttype
confi-guration.
The
pair
correlation functions are shown infigure
4and the
properties
of their firstpeak
aregiven
intable II. When the
germanium
concentrationincreases,
the
position
of the first maximum ischanging slightly
to smaller values of R and the
intensity
of thepeaks
decrease for an
alloy
richer ingermanium. Moreover,
we should remark that the first minimum becomes less
deep.
Theseproperties
show thatlong-range
order effects seem to
disappear
withincreasing germanium
concentration.FIG. 4. - Pair correlation functions for : a) AgO.858GeO.142’
b) Eutectic Ago.?s9Geo.24u c) AgO.402GeO.598’
The
comparison
with the work of Dutchak[15]
isdifficult as concerns the eutectic
composition
becausehe worked at a maximum
temperature
of 1 023 K.For all the
compositions,
we have calculated the interference functions in the Percus-Yevick appro- ximation[16] (see appendix 2).
The evaluated curves are shown infigure
5. The obtainedhard-sphere
FIG. 5. - Interference functions obtained by means of a Percus-
Yevick calculation at 1 123 K for : a) Ago,858Geo,142. b) Eutectic A90.759Geo.241- c) Ag0.402Ge0.598’
TABLE II
1442
diameters are
RG,
= 2.610A
andRAg
= 2.602A.
We find a
good agreement
between theposition
of themaxima and minima calculated and obtained
by experiments
for the 0.858 at% Ag alloy.
This agree- ment is not sogood
for the eutecticalloy
and very bad for the 0.402 at% Ag alloy.
This discordance may arise from the fact that the model chosen takes into accountonly
the size effect of theparticles.
In par-ticular,
it does not include the chemical interaction between the constituents. It thereforeprovides
uswith a
good representation
of asimple
metallic systemas has been shown in the case of pure
silver,
but itcannot take into account covalent
bonding
withdirectional effects.
On the basis of measurements of the three
alloys,
the
partial
interference functions are deduced fromequation (1) using
the concentration method.The
partial
interference functionSAgAg(k)
is repre-sented in the
figure
6together
with the calculatedSAgAg(k)
obtainedby
the Percus-Yevick method for the eutecticcomposition.
It should be noticed that avery
good
agreement exists between thepositions
andthe intensities of the maxima and the minima for these two interference functions.
Moreover, SAgAg(k)
is verysimilar to
S(k)
for pure silver(Fig. 1).
FIG. 6. - Partial interference function SAg-pg(k). Percus Yevick calculation deduced from an evaluation for the eutectic composi-
tion (full line). Calculated by means of concentration method
using experimental results near the eutectic composition (dotted line).
FIG. 7. - Partial interference function SGe-Glk). Experimental
curve of Isherwood et al. at 1 277 K (full line). Calculated by means
of concentration method using experimental results near the
eutectic composition (open circle).
The
partial
interference functionSGeGe(k), figure 7,
has all the characteristics of the structure factor of puregermanium. Moreover,
we obtained the resolution of the shoulder foundby
Hendus[17],
Krebs
[18],
Isherwood et al.[19],
Waseda et al.[20]
ina second
peak.
The firstpeak
ishigher
andsharper
asshown
by
the curve of Manaila et al.[19]
which ispresented
onfigure
7 withSGece(k).
5. Discussion. - The determination of the struc- ture factor of
liquid
silverby
neutron diffraction is in verygood
agreement with the Percus-Yevick eva-luation. This result is confirmed
by
the calculation of thepair
correlation function and coordination number which are wellrepresentative
of a compact system with pure metallic bonds. The distanceRi
of the firstpeak
ing(R)
and the nearestneighbours
number n,reported
on table I lead us to assume that local orderseems to be
unchanged
onmelting.
The interference functions of the
silver-germanium alloys
exhibit adecreasing
order as the concentration ingermanium
increases.It seems that in this range of concentration around the eutectic
composition
theordering
effect is due tometallic bonds for silver-silver
pairs
very similar to those of puresilver,
and bonds which recall the cova-lent
bonding
inliquid germanium
athigher
tempera-ture for
germanium-germanium pairs. Nothing
canbe concluded on the silver
germanium bonding;
moreover the
hypothesis
ofunchanged partial
inter-ference functions with concentration may introduce
uncertainty
on their calculation. The elaboration of acomplete
model of local order insilver-germanium alloys
needs a more accurateknowledge
of all thepartial
interference functions andpair partial
corre-lation functions. This studies
by
means of theisotopic
method are now undertaken.
6. Conclusion. - The structure factor and the
pair
correlation function of
liquid
silver at 1 323 K havebeen determined
by
means of neutron diffraction. The verygood
agreement of theS(k)
with a Percus-Yevick calculation and thehigh
value of the coordination number : 11. 93 exhibit a compact local order for silver in theliquid
state.Similar
experiments
On thesilver-germanium alloys
at 1 123 K have
permitted
us to determine the inter- ference functions. From theseresults,
we have deduced thepartial
interference functions for silver-silver andgermanium-germanium pairs
which are not far fromthose of pure elements. In addition the
present experi-
mental results could be used to obtain the
pair partial
correlation functions.
However,
in order to achieve thisresult,
it is necessary to check thevalidity
of thehypothesis
of invariance ofpartial
interference func- tions withcomposition.
This can be done with the aid of newexperiments using alloys
of variableisotopic
concentration of silver. As
suggested
in section1,
theseexperiments,
which are in progress, will be coveredby
asubsequent publication.
Acknowledgments.
- The authors wishespecially
to thank Dr. W. Knoll for his assistance
during
the
experimental
work at theLL.L.,
Dr. Y.Malmejac,
Mr. C.
Lemaignan
and Mr. J. P. Nollin for their collaboration inelaborating
thesamples
at the C.E.N.Grenoble.
Appendix
1. -By generalizing
a method of cal-culation used
by
Paalman andPings [21],
the inten-sities :
h(2 0), 12(2 0)
and/3(2 Ø)
can beexpressed
with the aid of the
equations :
IF
is the theoreticalscattering
of the furnacealone, Ic
is the theoreticalscattering
of the containeralone,
andIE
is the theoreticalscattering
of thesample
alone.The word theoretical
scattering
indicates that thebeam has not been attenuated
by
eitherscattering
orby
pureabsorption.
h
=intensity
scatteredby
the furnacealone, 12
=intensity
scatteredby
the furnace and thecontainer,
and13
=intensity
scatteredby
the furnaceplus
contai-ner
plus sample.
AF,F
is theabsorption
term due to the furnace for thescattering by
thefurnace ;
AF,FC
is theabsorption
term due to the furnace and to the container for thescattering
of thefurnace ;
Ac,FC
is theabsorption
term due to the furnace and to the container for thescattering
of thecontainer;
AF,FCE
is theabsorption
term due to thefurnace,
tothe container and the
sample
for thescattering
of thefurnace;
Ac,FCE
is theabsorption
term due to thefurnace,
tothe container and the
sample
for thescattering
of thecontainer ;
AE,FCE
is theabsorption
term due to thefurnace,
tothe container and the
sample
for thescattering
of thesample.
They
are definedby
thefollowing expressions :
where - J.lF, pc, /IE are the
absorption
coefficients of thefurnace,
container andsample (Fig. 8).
-
KF RF
andRF
are the outside and inside radius of the furnace.-
Kc Rc
andRc
are the outside and inside radius of the container.If we evaluate the distance
XF
coveredby
the neu-trons in the
furnace, according
tofigure
8FIG. 8. - Evaluation of the optic path in the absorption process.
1444
where
Xl
andX2
are smaller thanRc.
It caneasily
beshown that as far as
Rc « RF, exp( -
,uFXF)
isindependent
of R and T such that we obtain :thus :
in the same way :
The coefficients
AE,cE(2 0), Ac,cE(2 0), Ac,c(2 0)
andAF,F(2 0)
are evaluated on the basis of a computer cal- culation. The calculation of the coefficientsAF,Fc(2 0)
and
AF,FCE(2 0)
is morecomplicated
and has necessi-tated a
special
program. Theintensity IE(2 0)
scatter-ed
by
thesample
thus appears in thefollowing
form :Appendix
2. - For a two-component system, the P.Y.equation
may be written :where the
CaP(R)
are the direct correlation functions andUaP(R)
are thepair potentials.
The solutions for the
special
case ofhard-sphere
interactions between the
components
have beengiven
in detail
by
Lebowitz[16]
and discussedby
Ashcroftet al.
[22].
We repeat thesebriefly
in thefollowing :
where r
The ai,
bl, d,
a2,b2
coefficientsdepend only
on thediameters Q1
and U2
of the hardspheres,
thedensity
pl and P2 and the totalpacking
fraction for the mixture 11.We recall the definition
of 11 :
volume occupied by
hardspheres
11 =
volume total
The
Ca.p(k)
functions are evaluatedby
Fourier trans-form of the
Ca.p(R).
Thenby
means of the Ornstein- Zemikerelations,
theSa.p(k)
functions are deduced.References
[1] TOURAND, G. et BREUIL, M., Colloques Internationaux C.N.R.S. n° 205 Odeillo 1971 (édition CNRS) 1972, Paris.
[2] CINDA, International Atomic Energy Agency, Vol. I (1975).
[3] FABER, T. E., Introduction to the theory of liquid metals (Cam- bridge, University Press) 1972.
[4] Neutron Beam Facilities, I. L. L. High Flux Reactor (1973).
[5] TOURAND, G., J. Physique 34 (1973) 937.
[6] PLACZEK, G., Phys. Rev. 86 (1952) 377.
[7] YARNELL, J. L., KATZ, M. J., WENZEL, R. G. and KOENIG, H. S., Phys. Rev. A 7 (1973) 2130.
[8] BLECH, I. A. and AVERBACH, B. L., Phys. Rev. A 137 (1965)
1113.
[9] HULTGREN, R., DESAI, P. D., HAWKINS, D. T., GLEISER, M.
and KELLEY, K. K., Selected values of thermodynamic properties of binary alloys, American Society for metals,
Ohio, 1973.
[10] MARTIN-GARIN, L., GOMEZ, M., BEDON, P. and DESRE, P., J. Less. Common Met. 41 (1975) 65.
[11] WAGNER, C. N. J., OCKEN, H. and JOSHI, M. L., Z. Naturf.
20a (1965) 235.
[12] PFANNENSCHMID, O., Z. Naturf. 15a (1960) 603.
[13] WASEDA, Y. and OHTANI, M., Phys. Stat. Sol. b 62 (1974) 535.
[14] ASHCROFT, N. W. and LEKNER, J., Phys. Rev. 145 (1966) 83.
[15] DUTCHAK, Ya. I., Elektrokhim. Rasplavy (1974) 131.
[16] LEBOWITZ, J. L., Phys. Rev. 133 (1964) A 895.
[17] HENDUS, H., Z. Naturf. 2a (1947) 505.
[18] KREBS, H., LAZAREV, V. B. and WINKLER, L., Z. Anal. Allg.
Chem. 353 (1967) 277.
[19] ISHERWOOD, S. P., ORTON, B. R. and MÃNÃILÃ, R., J. Non- Cryst. Solids 8 (1972) 691.
[20] WASEDA, Y., SUZUKI, K., Z. Phys. B 20 (1975) 339.
[21] PAALMAN, H. H. and PINGS, C. J., J. Appl. Phys. 33 (1962)
2635.
[22] ASHCROFT, N. W. and LANGRETH, D. C., Phys. Rev. 156 (1967)
685.