300 Ionics 8 (2002)

Simple Method to Determine Electronic and Ionic Components of the

Conductivity in Mixed Conductors
A Review

R.A. Huggins
Faculty of Engineering, Christian-Albrechts-Universitiit
24143 Kiel, Germany

Abstract. In materials that are of interest for use as solid electrolytes it is important that charge
transport is predominantly related to ionic motion, with minimal minority electronic conduction. On
the other hand, a number of materials are mixed ionic and electronic conductors.
Several DC methods that have been developed over many years to evaluate the separate electronic
and ionic components of the total charge transport in solids. These include the Tubandt DC method,
the DC assymetric polarization technique sometimes called the Hebb-Wagner method, and the DC
open circuit potential method.
This paper presents a new method that is relatively simple, and could be advantageous in some
cases. It involves the employment and proper interpretation of low amplitude variable frequency
AC measurements, a general technique that is often called impedance spectroscopy. As an example,
the results of the use of this method on a mixed-conducting perovskite related to the well-known
LSGM, Lao.4Pro.4Sro.2Ino.sMgo.202. 8, are shown.

1. Introduction microstmcture of battery electrodes, and as fuel cell

Electrical charge can be transported within solids by the
net macroscopic motion of either electronic or ionic
species. In materials that are of interest for use as solid
electrolytes it is important that the charge transport is
predominantly related to ionic motion. Minority
electronic conduction is generally considered to be
deleterious. On the other hand, ionic charge transport is
generally not desired in materials such as electronic
semiconductors, for it can cause changes in the properties
of junctions as well as influencing the concentration of
traps and other features of the microstructure. Between
these two extremes are a number of materials that are
mixed ionic and electronic conductors. Such mixed-
conductors play important roles in a number of different
technologies, being employed as oxidation and hydro-
genation catalysts, as chemical separation membranes,
as reactants or mixed-conducting matrices in the
electrodes.
A number of methods have been developed to evaluate
the separate electronic and ionic components of the total
charge transport. Several of these will be briefly dis-
cussed. However, the primary purpose of this commu-
nication is to present another method that is relatively
simple, and could be advantageous in some cases. It
involves the employment and proper interpretation of low
amplitude variable frequency AC measurements, an ex-
perimental procedure that is often called impedance spec-
troscopy and employed for other purposes, such as the
evaluation of the ionic conductivity of solid electrolytes.
2. Transference Numbers of Individual Spe-
cies
If more than one species can carry charge in a solid
or liquid it is often of interest to know the relative
Ionics 8 (2002)
conductivities or inversely, the impedances of different
species. The parameter that is used to describe the con-
tributions of individual species to the transport of charge
is the transference number, sometimes called the Hittorf
transference number [1]. This is defined as the fraction of
the total electrical current that is carried by a particular
species when an electrical potential difference is imposed
upon the adjacent electrodes.
In the simple case that electrons and one type of ions
can move through such an electrochemical cell, we can
define the transference number of ions as ti, and electrons
as t e, where
t i = ii/(i i + ie)
and
te = ie/(i i + ie)
and ii and i e are their respective partial currents. We can
readily see that the sum of the transference numbers of all
mobile charge-carrying species is unity. In this case we
have:
ti+ te = 1
Instead of expressing transference numbers in terms of
currents, we can also write them in terms of the related
impedances, Zi and Z e. For the case of a material
containing these two mobile species, and the transport of
charge by the motion of the ions under the influence of an
applied voltage Eappl,
t i = (Eappl/Zi)/[Eappl/Z i + Eappl/Ze] = (Ze/Zi + Ze)
and likewise for electrons:
te = Z4(Zi + Z~)
Whereas these parameters are often thought of as pro-
perties of the electrolyte or mixed-conductor, in actual
experiments they can also be influenced by what happens
at the interfaces between the electrolyte or mixed-con-
ductor and the electrodes. They are only properties of the
electrolyte or mixed-conductor alone if there is no im-
pedance to the transfer of either ions or electrons across
301
the electrolyte/electrode interface or mixed-conductor/elec-
trode interface, or within the electrodes.
The transference number is also not a constant for a
given material, but instead, is dependent upon the che-
mical potentials of the constituent components within it.
Thus it is dependent upon the local composition and will
not be uniform throughout an electrolyte in a galvanic
cell in which the electrodes have different values of the
chemical potentials.
3. Tubandt DC M e t h o d
Many years ago Tubandt [2] introduced a DC method to
evaluate the ionic fraction of the total electrical con-
(l) ductivity in a solid. This involves the use of electrodes
that are not blocking to either ions or electrons, and the
imposition of a DC current of known time and mag-
nitude. Changes in the weights of the two electrodes are
(2) measured, one getting lighter, and the other heavier.
Assuming Faraday's law and knowledge of the atomic
weight of the transporting ionic species, this can be used
to determine the number of moles of material that were
ionically transferred. Comparison with the number of
Faradays of charge that passes through the system then
gives the value of the ionic transfer number, and thus also
(3) of the electronic transference number.
4. DC A s s y m e t r i c P o l a r i z a t i o n M e t h o d
A different method was introduced by Wagner [3] that can
be especially useful for the evaluation of low levels of
electronic conductivity in materials that are primarily
ionic conductors.
The general philosophy behind this type of measure-
ment involves the independent measurement of the current
carried by minority electronic species under conditions
(4) such that ionic transport is prevented. This involves the
use of one electrode that is reversible to both ions and
electrons, and another that is blocking to the mobile ionic
species. This is accomplished by using a electrode that
(5) does not contain atoms of the mobile ionic species, and
polarizing the cell such that they tend to move away from
that electrode into the electrolyte. Since the electrode
cannot supply those ions, the electrolyte becomes locally
starved, and ionic transport is prevented.
The polarity must be correct. If the mobile ionic
defects are positively charged, this electrode must be made
positive relative to the other electrode. If the mobile ionic
defects carry negative charges, this ionic defect-starving
302 lonics 8 (2002)
26

Total e
Current 2o Current 15

lO
Hole Current 1o . ~ ElTon Current

90.I
\, I
0.05
IIIIIII

O. I
Voltage = Vmi~ble -Reference
............... b
s

0.05
Electron Current . . . ~ . . ~ ~ s
-I0 Voltage I= Vmilble- Reference -Io

-20 9

930 '

Fig. 1. Dependence of the electron, hole and total currents
in a Hebb-Wagner experiment upon the relative potentials
of the reference and variable electrodes when the reference
electrode potential is where the electron and hole con-
centrations are equal.
Fig. 3. Dependence of the electron, hole and total currents
upon the relative potentials of the reference and variable
electrodes when the reference electrode potential is 0.06 V
positive of where the electron and hole concentrations are
equal.

electrode must be on the negative side of the experimental ItotaI = [(RTA)/(LF) - exp(-u)) + Crh(exp(u ) - 1)]
cell. (6)
As the other electrode is reversible to the mobile ionic where E is the voltage applied between the variable and
species, its composition and electrical potential define the reference electrodes, u = EF/RT, A and L are the area and
local composition and chemical potentials of the material length of the sample being measured, and cre and Ch are
(electrolyte) that is being measured. The steady state the partial conductivities of electrons and holes at the
current through the cell is evaluated as a function of the composition of the reversible reference electrode. F is the
voltage between this reference elecla'ode and the other Faraday constant, R the gas constant, and T the absolute
variable ion-blocking electrode. temperature.
Under these conditions the steady state current is given Usually, the contribution o f either electronic con-
by duction or hole conduction will dominate, and the domi-
nant species can be identified from the shape of the I/E
curve. It was shown recently [4] that the relative mag-
nitudes of the electron and hole currents, and thus the
Total Current shape of the I/E curve, which is due to the transport of
30
both electrons and holes, is dependent upon the potential
Current "'~'--
ron Current of the reference electrode. This can make the separate eva-
20
luation of these factors difficult. This feature is illustrated
in Figs. 1, 2 and 3, in which the potential of the reference
electrode is varied. It is readily evident that the shape of
0 I 0 05 "~' i 0,05 0.1 0.15 0,2 the data is highly dependent upon both the polarity of the
/ ,F
' /0 k electrodes and the potential of the reference electrode.
Hole Current Voltage = Variable - Reference
These factors are often neglected when such experiments
-20
are undertaken.
-30 One could use a variable-composition mixed conductor
as the reference electrode, and control its potential by the
-40
use of a simple double electrochemical cell configuration
such as that shown schematically in Fig. 4. In this way
Fig. 2. Dependence of the electron, hole and total currents the partial electronic and hole conductivities can be varied
upon the relative potentials of the reference and variable
electrodes when the reference electrode potential is 0.06 V over a relatively wide range in order to clearly separate
negative the electron and hole concentrations are equal. them.
Ionics 8 (2002) 303

t.O0

- - 0.75
I

g 0.50 - -u=(F/RT)E

0.25 I I I
0 5 I0 15 20
exp(-u)

Fig. 4. Double cell arrangement that can be used to vary the Fig. 6. Measurements of the minority electronic conduction
potential of the reference electrode in a Hebb-Wagner is yttrium-doped thorium oxide [5].
experiment [4].

Another way to separate these two factors was pre- 1. The voltage that is placed across the electrochemical
sented by Patterson [5], who showed that eq. (5) can be cell must not exceed the decomposition voltage of the
rearranged to give electrolyte that is being evaluated. If this is not the
case, electrolytic decomposition of the electrolyte will
Itotal/[exp(u ) - 1] = [(RTA/(LF)] [Onexp(-u) + (~p] (7) produce an additional ionic current.
2. The potentials of both the reference electrode and the
The electron and hole conductivities can be simply ionically-blocking variable electrode must remain
determined by plotting the experimental results as within the chemical potential range in which ionic
Itota]/[exp(u) - 1] versus exp(u). The value of Oe is defect species are dominant. What happens when this
calculated from the slope, and (~h from the intercept. is not the case has been discussed by Crouch-Baker
Two examples of the use of this method are shown in [7].
Figs. 5 and 6, from [5]. 3. The partial conductivities of the different species vary
In order to produce meaningful results by use of this with the potential, and thus the composition, of the
DC assymmetric polarization procedure, careful con- material being measured. This means that the results
sideration must be given to several additional factors. that will be obtained depend upon the compositional
range, over which the measurement is made. There are
20
two parts to this matter. One is the composition, and
therefore the potential, of the reference electrode. The
15
"P'i''~ " " other is the range of composition investigated by
o
:r,.
-
.-- .9.31/O-S(~-tcm - i )
variation of the potential difference between the
variable potential electrode and the reference electrode,
as well as the polarity of the cell.
, "ai~"~176 T.3s~.c 4. One must wait until steady state is obtained. Since
placing a voltage across the cell results in the im-
position of a gradient in the chemical potentials of the
o m I | I I components within it, there must be a corresponding
500 I000 1500 2000 2500 3000
exp(-.)
rearrangement of the defect species concentrations. A
substantial time may be required before this new equi-
Fig. 5. Replotted data for AgBr at 353 ~ and 372 *C
measured by Ilschner [6]. librium state is attained.
304
5. DC Open Circuit Potential Method
A voltage is expected if a difference in the chemical po-
tential of the electroactive species is imposed across an
electrolyte under open circuit conditions such that neither
chemical species nor electronic current can be transported
externally from one electrode to the other. Under those
conditions, the measured voltage Eout is equal to the ther-
modynamic voltage Eth given by
Eout = Eth = -(AGr)/(zq) (8)
where AGr is the change in Gibbs free energy of the
virtual reaction, z is the charge number of the mobile
species and q is the magnitude of the elemental charge. If,
on the other hand, there is electronic leakage through the
electrolyte, the measured voltage is given by
Eout = ti Eth = (-ti AGr)/(zq) (9)
where ti is the ionic transference number [8,9]. This
method requires that well-defined and known thermo-
dynamic conditions are maintained at each of the two
electrodes, which can sometimes be difficult to achieve.
It has recently been shown [10] that this result is also
readily obtained by the analysis of the electrical response
of a simple equivalent electrical circuit that represents the
properties of an electrochemical cell with different poten-
tials at the two electrodes, as is the case with batteries and
fuel cells. This is shown in Fig. 7, where Eth is the
thermodynamic voltage, Iout the output current, Zi the
internal impedance to the transport of ionic species, and
Ze the impedance related to internal electronic leakage
through the electrolyte. Under open circuit conditions lout
is zero. Impedances are used instead of resistances, as they
are often frequency, or time, dependent.
Zi
Eth Ze
E which has a finite resistance, R i this physical confi-
I~ I out
0 valent circuit shown in Fig. 9 at low frequencies. The
Fig. 7. General equivalent circuit of battery or fuel cell [10].
lonics 8 (2002)
Fig. 8. Simple experimental arrangement generally used for
variable frequency AC measurements to evaluate ionic trans-
port in a solid electrolyte.
From this simple circuit it can be seen that
Eout = [Ze/(Zi + Ze)] = ti Eth (10)
or
Eout = Eth (1 - te) (11)
6. Variable Frequency Small Signal AC
Method
Variable frequency AC methods are now often used to
evaluate the charge transport properties of potential solid
electrolytes. They involve the use of electronically-con-
ducting electrodes on two sides of the solid to be investi-
gated. The experimental arrangement generally looks like
that shown in Fig. 8.
It is generally assumed that the electrical current with-
in the solid phase between the electrodes is carried by
ionic species, but that they are blocked at the electro-
lyte/electrode interface, whereas the transport of electrons
is not. Both theoretical and experimental, considerations
have shown that such an interface between a solid elec-
trolyte, considered to be a purely ionic conductor, and a
purely electronic conductor can be simply modelled as a
parallel plate capacitor [11]. The excess ionic charge on
one side of the ionically-blocking interface is balanced by
excess electronic charge in the adjacent metal electrode.
Since charge passes from one electrolyte/electrode
interface to the other by passing through the electrolyte,
guration can be represented by a simple electrical equi-
capacitive properties of the two electrolyte/electrode inter-
faces are combined into a single capacitance, Cin t.
Ionics 8 (2002)
iat
Fig. 9. Simple equivalent circuit for the low frequency be-
havior of an electrochemical cell with an ionic conductor
between two ionically-blocking, but electronically-conduc-
ting, electrodes.
At higher frequencies, however, the experimental con-
figuration shown in Fig. 8 begins to exhibit the effects of
an additional geometrical capacitance due to the presence
of a material (the electrolyte) with a finite dielectric con-
stant between the two parallel metallic electrodes. This
parallel plate capacitance, Cgeo m, acts across the whole
configuration, and typically has very small values. Thus
the equivalent circuit becomes that shown in Fig. 10,
which is often called the "Debye circuit".
This equivalent circuit has the desirable characteristic
that it also corresponds closely to the physical arrange-
ment. Typical values of Cgeom and Cin t are about 10-12
and 10-6 F/cm 2, respectively.
The experimental results are typically expressed in
terms of a plot of the imaginary and real parts of the
impedance, or the imaginary and real parts of the ad-
mittance, against each other on the complex plane. Since
this topic may be unfamiliar to many readers, and one
often sees misinterpretation o f experimental data in the
literature, a brief introductory explanation is included
here.
7. Representation of the Properties of S i m p l e
Circuit Elements on the Complex Impedance
Plane.
The impedance of a resistance Z R is simply its resistance
R, which is in phase with an applied AC voltage signal.
o_ [ic ell 9 Ri
lilt
Fig. 10. Debye equivalent circuit corresponding to the phy-
sical configuration of a solid electrolyte between two
electronically-conducting, but ionically-blocking, elec-
trodes.
305
l ZR
R
Fig. 11. Impedance of a simple resistance on the complex
impedance plane. The value of ZR is independent of fre-
quency.
One the other hand, the impedance of a capacitance Zc is
out-of-phase with the applied AC signal, and is called the
reactance X, where
X = (-j)/((~C) (12)
and
c0 = 2~f (13)
The representation of these simple elements on the
complex impedance plane, sometimes called an "Argand
diagram", is illustrated in Figs. 11 and 12, in which the
negative value of the imaginary part of the impedance, the
reactance, is plotted versus the real part, the resistance
over a range of frequency.
If there is a resistance in series with a capacitance the
impedance of this combination ZRC is given by
ZRC = R + jX = R -j/(toC) (14)
R
Fig. 12. Impedance of a simple capacitance on the complex
impedance plane. The value of the reactance ZC of the capa-
citor varies with the frequency. Its absolute value
approaches zero as the frequency becomes larger. In the ab-
sence of a series resistance, its values lie along the ab-
scissa.
306
-X
R
" O
Fig. 13. Appearance of the impedance of a series combi-
nation of a resistance and a capacitance upon the complex
impedance plane.
and thus varies with the measurement frequency. The
complex impedance plot will appear as shown in Fig. 13.
On the other hand, the impedance of a parallel con-
figuration of a resistance and a capacitance, ZR-C has the
shape of a semicircle on the complex impedance plane, as
shown in Fig. 14.
One can readily understand this behavior qualitatively.
At very low frequencies the impedance of the capacitor is
very large. Thus essentially all the current flows through
the resistor and its properties dominate the behavior. At
very high frequencies the impedance of the capacitor be-
comes very small, so that it effectively shorts out the
resistor and the total impedance tends toward zero. The
two impedances are equal, and equal current flows through
the two legs, when the resistive component is one half of
ZR. This is the center of the circle, so that it is also the
value of the reactance of the capacitor. Thus
ZR_C = ZR/2 = (-j)/(o.,~C) (15)
and therefore
o~RC = 1 (16)
_!
ZR -C
R ~
Fig. 14. Appearance of a parallel combination of a resis-
tance and a capacitance upon the complex plane. The
direction in which the impedance varies with increasing
frequency is indicated by the arrow.
lonics 8 (2002)
O O
vw_
O
rl ra
Fig. 15. An example of two equivalent circuits that, by
modification of the component values, will have the same
frequency-dependent impedance. In this case it is easy to see
that R! is equal to r I + r2.
at the angular frequency corresponding to the top of the
semicircle. It is possible to determine the value of the
capacitance from this relation and the experimental data.
The interpretation of physical processes by the use of
alternating frequency methods and the interpretation of the
results in terms of equivalent circuits is not always
straightforward. In general, more than one equivalent
circuit can be devised that will describe the experimental
results. Figure 15 shows two circuits that have the same
form of response at all frequencies. One is essentially a
parallel arrangement of series units, and the other is a
series arrangement of parallel units. The values of the
component magnitudes are not the same in the two cases,
however.
It is most useful to use a circuit that most directly
reflects the physical arrangement and processes in the
experiment. For instance, parallel networks would ob-
viously not be appropriate for two physical processes that
are known to occur in sequence or in series. Nor would a
series configuration be appropriate for physical processes
that are known to occur in parallel.
In the simple case of measurements on a solid elec-
trolyte using an experimental configuration such as that
shown in Fig. 8, assuming that there is no significant
electronic conductivity in the solid electrolyte phase and
that the ionic species are completely blocked at the
electrolyte/electrode interfaces, the data can be readily
interpreted by use of the standard Debye circuit discussed
earlier and shown in Fig. 10. The corresponding complex
plane plot is shown in Fig. 16. If the temperature is
varied the time constants of the various processes change
and the experimentally observable portion of the total
frequency response (the frequency window) shifts.
Ionics 8 (2002) 307

-X
I
Increasing / /
-X
frequency
Rlo~c
R

Fig. 16. Schematic complex plane plot for the experimental

arrangement shown in Fig. 8 that corresponds to Debye
equivalent circuit shown in Fig. 10.
This is a combination of the behavior illustrated in
Figs. 13 and 14. At very low frequencies no current flows
through the geometric capacitance Cgeom, because of its
very small value, typically of the order of picofarads,
which gives it a very large impedance. Thus it is not
visible in the measurements. Therefore the data show a
"low frequency tail" characteristic of the series R i - Cint
configuration. At higher frequencies more and more
current moves through Cgeo m, so the data appear as
though the equivalent circuit becomes a parallel confi-
guration of Cgeom and R i.
The figure gets distorted, and the impedance does not
go all the way down to the real axis at intermediate
frequencies if the values of Cgeom and Cin t are not suffi-
ciently different from each other. This is shown in Figure
17. It generally results from the use of electrodes that
only poorly contact the solid electrolyte, leading to a re-
duced value ofCin t. The values ofCin t are typically 15 to
40 microfarads per cm 2 [12] for an interface with com-
plete contact between the electrolyte and the electro-
nically-conducting electrode. Poor contact can lead to
R r
Fig. 18. Complex impedance plane plot showing tipped
straight line at low frequencies.
A common experimental observation is that the low
frequency tail in complex impedance plots is not truly
vertical, but is a straight line inclined at a finite angle
from the vertical, at lower frequencies and/or higher tem-
peratures, as illustrated schematically in Fig. 18. An early
example of experimental data showing this behavior is
presented in Fig. 19. In the case of the corresponding
figure on the complex admittance plane the center of the
semicircular portion of the figure is lowered below the
real axis.
It was shown that this deviation from ideal Debye
circuit behavior relates to processes that occur at the solid
electrolyte/electrode interface [13], and that the electrical
response in that range of frequency can be represented by
an impedance of the form
Z = A0J- a - jBto - a (17)
- X (ohm)

much smaller values. 8x I0s .01

I /
~FrOelquenci
in Hz
se
6 x 10S

~ 1,00E+03

"~ e,OOE.02 ....l~'e~ng 4x l0 S

.03
6,00E+02

4.00E+02 .OS
2x I0 s .07
2.00E*02

O.OOE+O0
O,OOE 2,00E 4,00E 6,00E 8,00E 1,00E 1.20E
..r02 +02 --02 +02 +03 +03 I 1
Real Impedance Z x 10s 4 x 10S 6 x 105
- R (ohm)
Fig. 17. Typical changes in the complex plane figure if the Fig. 19. Low frequency portion of the expenmenta~ data
values of the two capacitances are not sufficiently different measured on 13-PbF2 at 138 ~ [15].
from each other.
308
that acts in series with the bulk ionic impedance.
This is different from the behavior of a completely
blocking electrode interface, that is assumed to act like a
frequency-independent electrical double layer capacitance,
so that the low frequency tail should be vertical, as shown
in Fig. 16.
If the electroactive species is not blocked at the
interface, but diffuses into or out of one or both of the
electrodes as a neutral reaction product according to Fick's
second law the low frequency data will fall onto a straight
line with a slope of 45 degrees, so the value of o~ in that
case will be 1/2 if the interface is planar. This involves
the introduction of a so-called Warburg admittance or
impedance into the equivalent circuit. This has been
treated fully elsewhere [14], and since it is not directly
related to the subject of this paper, it will not be further
discussed here.
8. Influence of Electronic Leakage Through
The Ionic C o n d u c t o r
If the material between the electrodes is a mixed con-
ductor, with both ionic and electronic contributions to
charge transfer, the equivalent circuit must be modified to
include an electronic current path in parallel with the
ionic current path. This is shown in Fig. 20.
If the electronic resistance is much lower than the
ionic resistance the corresponding complex plane plot is
shown in Fig. 21. Because there is no blockage of the
charge transport due to electronic species, there is no in-
terface capacitance, and thus no capacitive tail at very low
frequencies. Also, the electronic current will be much
larger than the ionic current, even at frequencies that are
sufficiently high that the interracial capacitance is shorted
out.
I1..
,,
! -X
Int
Fig. 20. Equivalent circuit for a Debye system with elec-
tronic current, as well as ionic current, through the elec-
trolyte.
Ionics 8 (2002)
T
-X
Frequenc~
R t-ltnx,ron~
R
Fig. 21. Complex impedance plane figure for the case of an
ionic conductor with dominant electronic leakage.
The semicircle in this figure is only due to the
interaction of the electronic resistance and the geometrical
capacitance. It has nothing to do with the ionic resistance
or the interfacial capacitance. Because the geometrical
capacitance is so small, if the electronic resistance is also
small, the frequency range o f the data in the semicircle
will be extremely high. This can be seen by evaluation of
the frequency at the top of the semicircle, where
O)ReCgeo m = 1. (18)
As a result, experiments on materials that are pre-
dominantly electronic conductors often do not show the
semicircular part of the overall data, but only a single
point at the low frequency end of the figure.
9. Case in Which Both Ionic and E l e c t r o n i c
Transport Are Significant
If there is a substantial amount of ionic transport, as well
as an electronic shunt at low frequencies, the complex
plane plot will look like that shown in Fig. 22. There arc
T
RI R2
R
Fig. 22. Complex impedance plane plot for the case that
both ionic and electronic transport are significant.
Ionics 8 (2002)
two semicircles, representing the parallel combination of
the electronic resistance and the geometrical capacitance at
low frequencies, and the parallel combination of the ionic
resistance and the geometrical capacitance at the higher
frequencies. There is no low frequency tail, as the
interfacial capacitance is shunted by the electronic current
at low frequencies, as discussed above.
There are two intercepts on the resistance axis,
signated as RI and R2. The one at low frequencies, R2, is
the same as the DC value, and thus is merely the elec-
tronic resistance, Re. At higher frequencies the impedance
due to the relatively large interfacial capacitance becomes
insignificant and is not observed in the measurements.
The equivalent circuit of Fig. 20 then becomes a parallel
arrangement with three legs, the geometric capacitance
and the two resistances, which are in a parallel
configuration. This leads to the second semicircle in the
impedance plane. The resulting intercept on the resistance
axis at the higher frequency, R 1, is a combination of the
two parallel resistances, Ri and Re. Thus
I/R1 = l/Ri + 1/Re (19)
which can be rewritten to give
R 1 = (ReRi)/(R e + Ri) (20)
The ionic and electronic fractions of the total con-
ductance, and thus of the conductivity, can be found.
The ionic conductance Gi is
G i = 1/R i = I/R 1 - 1/Re = (R 2 - R1)/(R1R2) (21)
and the electronic conductance Ge is
Ge = 1/Re = l/R2 (22)
The total conductance Gtotal is
GtotaI = G i + G e = (R 2 - R1)/(R1R2) + 1/R 2 = 1/R 1 (23)
The transference numbers of the ionic and electronic
species are equal to their respective partial conductances:
ti = Gi/Gtotal = (R2 - R1)/R2 (24)
and
309
te = Ge/Gtota1 = RI/R 2 (25)
Thus it can be seen that the transference numbers of
the ionic and electronic species can be obtained by simple
evaluation of the relative values of the two intercepts on
the real axis in a complex impedance plane plot.
If the ionic transference number is relatively large, and
the electronic transference number is relatively small, as
is desired for useful electrolytes, the right hand, lower
frequency, semicircle is considerably larger than the
higher frequency semicircle. In the extreme case of very
little electronic conduction RI approaches zero. On the
other hand, if the electronic conduction is dominant, the
low frequency semicircle becomes quite small, and Rl
approaches R2. There is no low frequency capacitive tail
on the impedance plot in either case.
10. Influence of an Additional Impedance D u e
to Transverse Internal Interfaces
Grain boundaries and interphase boundaries in poly-
crystalline materials can have a significant influence upon
their observable transport properties. In the case that
transport along such internal interfaces is faster than that
through the bulk crystals in the microstructure, the pre-
sence of such features oriented in the direction of the
charge flux can result in an enhancement of the related
species transport. Transverse interfaces of this type will
have little effect. On the other hand, if species transport
in such boundaries is slower than in the crystalline por-
tions of the microstructure, transport along them will
have little effect. However, high impedance grain boun-
daries or interphase boundaries that are oriented transverse
to the charge flux direction can act to impede transport
processes, and can introduce additional resistive and capa-
citive effects into the measurements.
If ionic transport through the solid is impeded by the
presence of internal interfaces, e.g. grain boundaries,
across which species must travel, the additional ionic im-
pedance must act in series with the bulk ionic resistance.
This partial blocking of the ionic flux will also typically
have an associated parallel local capacitance similar to the
capacitive effect at an ionically blocking exterior elec-
trode. This additional feature causes the Debye equivalent
circuit shown in Fig. 10 to be modified to become that
shown in Fig. 23.
This circuit thus has two parallel R/C configurations,
so that there should be two semicircles on the complex
impedance plane figure. It will look like the one shown
310 lonics 8 (2002)

[L
IIc. lifo e- R! !

Fig. 23. Equivalent circuit containing a parallel R/C com-
bination, Rg b and Cg b, in series with the ionic resistance
due to the presence of transverse internal interfaces, e.g.
grain boundaries, that impede ionic transport.
Re
Fig. 25. Schematic equivalent circuit representing the same
physical processes as that shown in Fig. 23, but also
including the effect of electronic leakage.

in Fig. 22, except that there is a low frequency tail. This verse interface impedance, and also some electronic con-
is seen in Fig. 24. ductivity, we must make a further modification of the
However, in this case the meanings of R 1 and and R 2 equivalent circuit. This simply involves putting an elec-
are different. At low frequencies the impedance of Cg is tronic resistance in parallel with the ionic conduction
very large, and the behavior is determined by the series path. The result is shown in Fig. 25.
arrangement of Ri, Rg b and Cin t. This produces the tail, In this case, as well as any other that has significant
and the value of R2 is thus (Ri + Rgb). electronic charge transport, there will be no low frequency
At high frequencies the impedance of the grain boun- tail in the complex impedance plane figure. Instead, there
dary capacitance approaches zero, so that it shorts out the will be a semicircle at very low frequencies due to the
grain boundary resistance. Thus the circuit becomes the parallel action of Re and Cg. But there will also be two
same as the Debye circuit, and the behavior is dominated additional semicircles, as there are two additional parallel
by the parallel arrangement of R i and Cgeom. R/C combinations that can contribute to the experimental
Thus when there is an internal interface impedance in observations, Rgb and Cgb, and R i and Cgeom.
addition to the bulk ionic impedance and no appreciable The result is that there will, in principle, be three
electronic conduction the value of R1 is that of Ri and the semicircles in the complex impedance plane figure, as
value of R 2 becomes that of Ri + Rgb. illustrated schematically in Fig. 26. The two at the higher
frequencies reflect the influence of the ionic transport part
11. Behavior When There is Internal Trans- of the system.
verse Interface Impedance as Well as Partial The values of the intercepts on the real impedance axis
Electronic Conduction can now be identified. The one at the lowest frequency,
If a material has these same general properties, ionic R3, is equal to the electronic shunt resistance R e. The one
transport through the bulk crystallites and some trans- at the next higher frequency range is due to the parallel

-X

-X

Rl R2 R3
R~ R2 R
R IP Fig. 26. Complex plane plot for the case of electronic
leakage as well as ionic transport through the bulk with an
Fig. 24. Schematic complex impedance plane figure corres- additional ionic impedance in series due to transverse inter-
ponding to the equivalent circuit shown in Fig. 23. nal interfaces.
Ionics 8 (2002) 311

Table 1. Interpretation of the Intercepts on the Real Axis.

Fype of conduction R1 R2 R3
Ionic conduction only Ri
Electronic conduction o n l y Re
9 I
Ionic with internal interface Ri R i +Rgb
impedance
Ionic with electronic shunt ReRi/(Re+Ri) Re
Ionic + interface impedance ReRi/(Re+Ri) (Re)(Ri+Rgb)/(Re+Ri+Rgb) Re
elect9 shunt

Table 2. Evaluation of the Different Resistances.

Type of Conduction I Re [ Ri Rgb

Ionic conduction only R1
Electronic conduction only R!
Ionic with internal interface impedance R1 R2 - R 1
Ionic with electronic shunt R2 R2RI/(R2-R1)
Ionic + interface impedance, elect, shunt R3 R3RI/(R3-R1) xxxxxx

Table 3. Transference Numbers.

Type of conduction Ionic Electronic

transference transference
number number
Ionic conduction only 1.0 0.0
Electronic conduction only 0.0 1.0
Ionic with electronic shunt (R2 - RI) / R 2 R 1 / R2
Ionic + interface impedance, (R 3 - R2) / R 3 R2 / R3
electronic shunt

combination o f R e and the sum o f R i and Rg b. The re- R 1 = (ReRi)/(Re + R i) (28)

sistance of this parallel combination can be found from
The transference numbers can now be found by first
I / R 2 = 1/Re + 1/(Ri + Rgb) (26) calculating the conductances, as was done earlier for the
simpler case.
Thus R 2 is given by The ionic conductance is

Re (Ri + R g b ) G i = 1/Rioni c = I/(R i + Rgb) (29)

R2 = (27)
R e + R i +Rgb
The electronic conductance is
and the one at the highest frequencies is due to the parallel
combination o f R i and Re, as Rg b is shunted out by the G e = 1/R e = I/R 2 - 1/(R i + Rgb) (3o)
grain boundary capacitance Cg b. As before, R1 can be
found to be And the total conductance is
312 Ionics 8 (2002)

Gtotal = Gi + Ge = 1/(Ri + Rgb) + 1/Re = l/R2 (31) Table4. Results of Experiments on La0.4Pr0.4Sr0. 2-
In0.8Mg0.202.8.
The ionic transference number can be found from: Temp(*C) ] R 1 ] R2 J Ri I Re ] ti [ te
160 6000 18000 9000 8000 0.67 0.33
t i = Gi/Gtota I = 1/(R i + Rgb) R 2 195 2500 5900 4338 5900 0.58 0.42
= (1/R 2 - 1/Re) R 2 = (R 3 - R2)/R 3 (32) 225 1875 3750 3750 3750 0.50 0.50

And the electronic transference number can be found

from: the data on the real impedance axis. These results ate
summarized in Table 3.
te = Ge/Gtotal
= (Ri + Rgb)/(Re + Ri + Rgb) = R2/R3 (33) 12. An Example
Experimental results on an Indium-containing analog of
Thus it can be seen that the presence of an electronic the well-known perovskite solid electrolyte LSGM,
leak through the electrolyte produces important changes Lao.4Pro.4Sro.2Ino.8Mgo.202. 8, at three different tempera-
in experimentally observed complex impedance plots tures are shown in Fig. 27, and the resultant data are
obtained from small signal AC measurements using shown in Table 4.
ionically-blocking electrodes. This electronic shunt leads
not only to a difference in the shape of the complex plane 13. Summary
figures, with no low frequency tail, but also necessitates A new, relatively simple, method based on impedance
different interpretations of the data. The relationships bet- spectroscopy using ionically blocking electrodes has been
ween the values of the intercepts of the data with the real presented that can be useful in interpreting experimental
impedance axis and the specific resistances within the data on mixed conductors when there is not a large
sample for several cases are given in Tables 1 and 2. difference between the ionic and electronic impedances.
The values of the ionic and electronic transference It does not require high ionic conductivity and does
numbers in the presence of transverse internal interfaces not involve passing a significant amount of material
in addition to electronic leakage are analogous to those for through the sample. Nor does it require the imposition of
the simpler case without the additional internal ionic im- different thermodynamic conditions upon the two sides of
pedance. In both cases the respective transference numbers the sample and waiting for slow equilibration of species
can be obtained merely by observation of the intercepts of within it. The whole sample can be surrounded by the
i | i i i i i same atmosphere, so data can be obtained as a function of
the thermodynamic environment.
-17500
One can easily recognize the presence of a significant
- 15000 amount of electronic leakage through the sample by the
9 160 *C
absence of a capacitive tail in complex impedance plots at
-~25oo " 195 *C
low frequencies. When that is the case, evaluation of the
"~ o 225 *C
E -loooo relative values of the intercepts upon the real impedance
t-
O axis provides information about the ionic and electronic
1"4 -7500
resistances, as well as their respective transference
-5000 9 9 9 9 9 9 9 9 numbers.
When substantial electronic conduction is present, the
meanings of these intercept values are distinctly dift~rent
from when the charge transport is only ionic. This has
0 2 5 0 0 50o0 7500 10000 12500 15000 17500
been shown for the situation in which there is no appre-
Z~(ohms) ciable impedance due to the presence of transverse internal
Fig. 27. Plot of the frequency dependence of the complex
interfaces, and also for the case in which internal, par-
impedance of La0.4Pr0.4Sr0.2In0.8Mg0.202.8 at several tem-
peratures. tially blocking, transverse interfaces are present.
Ionics 8 (2002) 313

In contrast to resistance measurements, and conduc- [5] J.W. Patterson, E.A. Bogren, and R.A. Rapp, J.
tivity values that result from them, transference number Electrochem. Soc. 114, 752 (1967).
values are relative, and thus independent of sample dimen- [6] B. Ilschner, J. Chem. Phys. 28, 1109 (1958).
sions. Therefore it is possible to obtain useful trans- [7] S. Crouch-Baker, Solid State Ionics 45, 101
ference number information using this AC method even (1991).
on samples containing significant amounts of poro- [8] V.B. Tare and H. Schmalzried, Z. physik. Chem.
sity. NF 43, 30 (1964).
One the other hand, it is important to recognize the [9] C. Wagner, Adv. in Electrochem. and Electrochem.
assumption that the electrodes are completely blocking to Eng. 4, 1 (1966).
ionic species. If this is not the case, the apparent value of [10] R.A. Huggins, Ionics 5, 269 (1999).
the transference number for the electronic species will be [11] D.O. Raleigh, Electroanal. Chem. 6, 87 (1972).
too high. [12] B.E. Conway, J. Electrochem. Soc. 138, 1539
(1991)
14. Acknowledgement [13] I.D. Raistrick, C. Ho, and R.A. Huggins, J. Elec-
The helpful assistance of V. Thangadurai and W.F. Chu trochem. Soc. 123, 1469 (1976).
is gratefully acknowledged. [14] C. Ho, I.D. Raistrick, and R.A. Huggins, J. Elec-
trochem. Soc. 127, 343 (1980).
15. References [15] I.D. Raistrick, C. Ho, Y.W. Hu, and R.A.
[1] W. Hittorf, Ann. Physik 106, 543 (1859). Huggins, J. Electroanal. Chem. 77, 319 (1977).
[2] C. Tubandt, Z. Electrochem. 26, 338 (1920).
[3] C. Wagner, in: 7th Meeting of the International
Paper presented at the 8th EuroConference on lonics,
Committee of Electrochemical Thermodynamics and
Carvoeiro, Algarve, Portugal, Sept. 16 - 22, 2001.
Kinetics, Butterworths, London, 1956.
[4] R.A. Huggins, Solid State Ionics 143, 3 (2001). Manuscript rec. Nov. I, 2001; acc. May 2, 2002.