Académique Documents
Professionnel Documents
Culture Documents
JOURNAL DE PHYSIQUE
Resume. - Nous avons mesure la conductivite ionique de AgCI, fortement dopC par Cdz+
(0 a 5 000 ppm). Dans le domaine de temperature 250-400 K, nous avons Ctudie la solubilite de
Cd++ dans AgCI. Pour la discussion des resultats obtenus en mesurant la conductivite ionique
au-dessous de 240 K et pour expliqucr les energies d'Arrhenius tres Clevees dans ce domaine de
temperature, nous avons propose les deux modeles suivants : 1) Association des dipdles aux
agregats plus grands. 2) Precipitations d'une superstructure : CdCI2, 6 AgCI.
Abstract. - Ionic conductivity measurements in heavily doped AgCI : Cd crystals (from 0
to 5 000 ppni) were carried out. A solubility curve of Cd? in AgCl for the temperature range
250 K to 400 K is presented. Very high Arrhenius energies were measured below 240 K. A dipole
association model and a superstructure model are discussed to explain the obtained conductivity
data. I t is shown that the experimental results can appropriately be described by the assumption
of the precipitation of this superstructure (CdC12.6 AgCI).
evaporated Ag-electrodes
1. p l a t e l e t s
/I
quarzglas
2 . Bridgman crystals
,evaporated
Ag-electrodes
2. Experimental. - 2 . 1 CRYSTALPREPARATION.
T w o different shapes of crystals were used in the
experiments (Fig. 1) :
a) Plate-like crystals about 200 11 thick, individually
doped u p to 5 000 ppm C d Z + and grown by the
sandwich method. This method was introduced by
Clark and Mitchell [6] and improved by Zorgiebel
and Wendnagel [S].
b) Discs of about 3 nim thickness cut from a cylindrical crystal of 15 mni in diameter grown by the
Bridgman method.
The second were used for comparison and to
exclude surface ell'ects. The impurity concentration
of the Bridgmnn-samples was determined by the
distribution coefficient [9].
T o achieve smooth surfaces, all samples were polished
with KCN
01- NaZS,O, - solutions. T o avoid
-
electrical resistance
with an ohmmeter
an input resistance
3 to 10 '%, dependent
C9-416
p =T eipj-
$1
I : intrinsic region
I1 : vacancy region
111 : dipole association region
1V : (( precipitation region ))
El = H F / -I-~ UO
EII = Un
EIII = Un
UA/~
Erv = UiIi
Ul,
+
+
Regions 111 alld IV are liere of special interest, especially because region I V is assumed to be a rather
unknown (( precipitation region D.
4. Results. - 4 . 1 DESCRIPTION
OF THE CONDUCPLOTS.- Figure 3 and figure 4 show the Arrhenius plots of the two types of crystals with various
impurity concentrations. Let us first consider the
curves of the (( pure )) crystals i n both graphs. We can
distinguish the regions I, 11 and 111. It is well known
from theory e. g. [ I 11, [I21 that the transition temperature between region I and I1 depends on the
impurity concentration, namely the lower the transition temperature the lower the impurity concentration. Consequently the undoped Bridgman crystals
are (( purer )) than the platelets. In the other curves
TIVITY
- (
exp
(3)
region
1.. Ag CI "pure"
1 1
2.0
4 CI 50ppm Cd
+
3.0 Ag C l * 500ppm C d
L.fiAgCI
'5.4
5.0
4.6
4.2
5 0 0 0 ppm Cd
3.8
3.4
3.0
2.6
2.2
1000
c-
C9-4 1 7
Miiller
(1965)
Abbink
and
Martin
(1966)
Corish
and
Jacobs
(1972)
Present
work
0.785
0.35
0.27
0.725
0.27
0.47
0.74
0.29
0.29
0.78
0.30
0.30
El (eV)
U , (eV)
U , (eV)
FIG.4.
x).
' 0.15.
C9-418
( x , = concentration of dipoles), K A ( T ) is tlie equilibrium constant of the reaction for the association
of dipoles :
- z2 e2 X
2 &kT(1 + XR)
I,
<5
x lo4 cm,
100 s the
The diffusion length being extremely small, the precipitation from a homogeneous distribution of Cd2+-ions
turns out to be very improbable. These inhomogenities
may be due to
1) locally enhanced concentration of two or more
dipoles which then will form dimers or trimers ;
2) dipoles localized near the surface of precipitations.
In both cases only short diffusion lengths are needed,
and each of the two ways mentioned above may be
used to interprete tlie slope in region IV.
5 . 2 DIPOLEASSOCIATION MODEL.- The dipole
association mechanism is illustrated in figure 7 [13],
[14]. The association energies U D for dimers and U ,
for trimers were roughly calculated by a simple Coulomb interaction taking into account the dielectric
constant of AgCI.
260
2i0
300
3$0
340
3k0
3b0
400
(zi =
I for Cd-ions ; zi = - 1 for vacancies ;
r i j = distance between charges i, j ; E = 12.5).
(100)
impurity
C9-419
of AUD and AU, are very small. Therefore the measured high Arrhenius energies of region 1V cannot
be explained by the dipole association model.
5 . 3 SUPERSTRUCTURE
MODEL. - It is reasonable to
discuss a superstructure CdCl, . 6 AgCl with twice
the lattice constant of the host lattice (Fig. 8).
(100)
- vacancy - dipole
dimer
trimer
FIG.8.
+-
(1 1)
X* = K D ( T )=
xD
( x , , xT
Thence
exp -
$1
0.90 eV .
0.52 e\'
By this simple consideration it is seen that the activation energies for taking off free vacancies from a
superstructure particle depends on the size and the
remaining charge of the precipitated state. By means
of this model i n comparison witti the dipole association model the measured high and different Arrhenius
2R
C9-420
References
[I] KOCH, E. and WAGNER,
C., Z . P11ys. Cller?~ieB 38 (1937)
295.
121 EBERT,I. and TELTOW,J., A I I I I ~der
I I Pllys. 5 (1949) 63, 71.
P.,, P I I ~ sSrar.
.
Sol. 12 (1965) 775.
[3] M ~ L L E R
[4] ABBINK,H. C. and MARTIN,
D. S., J. Phys. & Cl~c,r,~.
Solids
27 (1966) 205.
P. W. M., J . P11y.s. & Chern. Solids
[5] CORISH,J. and JACOBS,
33 (1972) 1799.
J. W., J. Photogr.
[6] CLARK,P. V. MC D. and MITCHELL,
Sci. 4 (1956) 1.
F. et al., Z. Angew. Plzys. 30 (1970) 316.
[7] ZORGIEBEL,
[8] GRANZER,F., HENIG, G., SCHOTT,U., DARDAT,K.,
WENDNAGEL,
Th., HAASE.G., Z ~ ~ R G I E BF.,
E L , in :
(( 8th Int. Conference on Nuclear Photography and
Visual Detectors, Bukarest (1972) D.
[9] SCHILDKNECHT,
H., (( Zo~~er~schn~elzeiz
)) (Verlag Chemie,
Weinheim) 1964.
DISCUSSION
A. KESSLER.
- If your crystals were not in perfect
equilibrium, one could expect some hysteresis in the
data goinp up and down with temperature. Have
you observed something of this kind ?
K. ZIEROLD.
-Yes. By cooling from room temperature or any other temperature in region I11 the
conductivity is lower than by heating. The contrary
effect is observed by cooling (or quenching) from the
intrinsic region.
R. CAPELLETTI.
- Which is the heating rate during
the ionic conductivity measurements ?
If it is high you are actually measuring non-thermodynamical equilibrium situation and the activation
energy in region IV experimentally obtained is expected to be higher than the real one.
Generally the accomplishment of thermodynamical
equilibrium between simple defects and precipitated
phase requires long time, for instance in alkali halides
doped with divalent cation impurities.