You are on page 1of 12

8

Nanoionics

Joachim Maier
Max Planck Institute for
Solid State Research

8.1 Introduction: Significance of Ion Conduction .................................................................... 8-1


8.2 Ionic Charge Carriers: Concentrations and Mobilities .....................................................8-2
8.3 Ionic Charge Carrier Distribution at Interfaces and Conductivity Effects ....................8-5
8.4 Mesoscopic Effects ...................................................................................................................8-8
8.5 Consequences of Curvature for Nanoionics...................................................................... 8-10
8.6 Conclusions............................................................................................................................. 8-11
References........................................................................................................................................... 8-11

8.1 Introduction: Significance


of Ion Conduction
While nanoelectronics refers to electronic transport and storage
phenomena on the nanoscale, nanoionics referson that same
scaleto ionic transport and storage phenomena. At interfaces,
and particular in confined systems, exciting ionic phenomena are
observed that indeed justify the use of this term (Maier 2005b).
Ion motion is no less significant for processes in nature and
technology than motion of electrons. Well-known is the role of
ion transport in liquid or semiliquid systems, a striking example being offered by biology in terms of nerve propagation. But
also as far as solids are concerned, the role of ion transport is
of paramount significance. All mass transport phenomena in
ionically bound solids require ion transport, usually in the form
of simultaneous transport of ions and electrons or of different
ions. Beyond that, there is a whole class of applications, typically energy-related applications, for which the mobile ions are
indispensable and their role cannot be taken over by electronics. Such devices include batteries and fuel cells with the help of
which electrical energy can be stored or just converted to chemical energy. Related applications are various types of chemical
sensors, chemical fi lters, or recently described resistive switches.
In this case, basically chemical information is transformed into
physical information.
Let us discuss some electrochemical applications based on
oxygen ion conductors such as Y-doped zirconia. Two different
oxygen partial pressures on the two sides of this oxide ceramic
generate a cell voltage that can be used to detect oxygen partial
pressures once the value on one side is known, or even to control that partial pressure as it is done in modern automobiles. If
one works with very reducing gases (e.g., hydrogen) on one side
(what corresponds to a low oxygen partial pressure), while having, for example, air on the outer side, the principle of a fuel cell

is realized; this case is indicated in Figure 8.1a. In this way, H2 is


electrochemically converted to H2O. As the electrical energy can
be directly used without thermal detour, Carnots efficiency does
not apply and high theoretical efficiencies can be expected. As
the solid electrolytes can typically be used at high temperatures,
gases such as hydrocarbons can be converted quickly enough.
If material problems that are connected with high temperatures are to be avoided, it is well advised to use proton conductors,
which are mobile enough at moderate temperatures. Then the
direction of the mass transport is changed and H2O is produced
on the air side where it does not pollute the fuel.
Rather than just conversion, cation conductors such as
Li-conductors allow for energy storage. In the same way as the
cell voltage in the previous examples is given by the difference in
the chemical potential of oxygen or hydrogen (partial pressure
of oxygen or hydrogen), the Li-potential difference is exploited
here. Unlike in previous examples, the decisive component (Li)
is accommodated in a solid phase (see Figure 8.1b). This provides
the possibility to efficiently store electrical energy while in the
above examples, it was rather the transformation of energy that
was important. The low weight of Li and its high electronegativity guarantee high energy per weight.
As indicated in Figure 8.1b, Li storage (i.e., Li+ and e) requires
both ionic and electronic conductivities. Mass transport enabled
by this mode is also exploited in chemical fi lters. Oxygen, for
example, selectively permeates through oxides, which exhibit
both O2 and e conductivities. Very related applications are gas
storage applications such as for H2 provided by polar hydrides
(here it is the simultaneous motion of hydrogen ions and electrons) or equilibrium conductivity sensors (conductivity effects
on varied stoichiometries as response to varied partial pressures). This possibility of having both types of conductivities in
the solid state enables gas fi ltering by mixed conducting permeation membranes.
8-1

8-2

Handbook of Nanophysics: Principles and Methods

Chemical

Energy
information

Physical

Energy
information

2e

O2
O2

H2

(a) Fuel cell

Electrolysis
e

Li+

Li+
Li+

e
(b)

e
Li-based battery

FIGURE 8.1 A few selected electrochemical applications (cf. text).

Moreover, the mixed conductor represents the general case of


an electrical conductor from which the pure ion conductor and
the pure electron conductor follow as special cases, making it a
master material of fundamental importance. We will refer to the
mixed conductor quite frequently in the following discussion.

8.2 Ionic Charge Carriers:


Concentrations and Mobilities
The top row of Figure 8.2 identifies typical ionic charge carriers in crystals, namely, particles in interstitial sites or vacancies
(Wagner and Schottky 1930), while the bottom row refers to
electronic carriers (in a localized picture).

M+ X M+ X

M+ X M+ X

M+ X M + X

The ionic excitation of an ion from its regular site to an interstitial site leaving behind a vacancy (Frenkel disorder) is analogous to an excitation of an electron from the valence band to the
conduction band, leaving a hole in the valence band. In many
cases, the electronic carriers can be connected with different
valence states (in the absence of substantial hybridization). For
example, in a component such as silver chloride, a neutral silver
on an Ag+ site is equivalent to an electron in the conduction band
while a missing electron on a Cl (i.e., neutral Cl) corresponds to
a hole in the conduction band. Charge delocalization (without
mixing Cl and Ag orbitals) means that the varied valence states
do refer to the ensemble of cations or anions rather than to a specific identifiable ion.
If we refer to a tiny energy interval and consider the statistics
in a sample of constant number of lattice sites, we have to refer
to FermiDirac statistics in both cases. In the case of electrons,
it is Paulis principle that excludes two electrons in the same
state, and in the case of ions, it is the restriction that two ions
cannot occupy the same lattice site (which after all is, of course,
also a quantum mechanical effect) (Wagner and Schottky 1930;
Kirchheim 1988; Maier 2004b, 2005a).
Even though the density of states is very differentin the case
of electrons we may have delocalization and parabolic density of
states, while in the case of ions we typically face sharp energies
for interstitial sitesthe statistics concerning the entire energy
range of interest is very similar as long as the gap (standard free
energy of formation) is sufficiently large. The chemical potential
of ions and electrons will then follow a Boltzmann distribution.
Figure 8.3 shows how a Boltzmann distribution results for different cases irrespective of the details concerning nature and
concentration of energy levels.
Figure 8.4 refers to the ionic and electronic excitations in the
energy level picture. In all cases, these energy levels are standard
electrochemical potentials, while Fermi/Frenkel levels are full electrochemical potentials including also configurational terms. (One
may use the term Frenkel level to emphasize the parallelity of the

MM Xx MM Xx
Xx

VM'

Xx MM

X M+ X M+

M + X M + X
M+
X M+ X M+

MM Xx MM Xx
Mi
Xx MM Xx MM

M+ X M+ X

M+ X M+ X

MM Xx MM Xx

X M+ X M+

X Mo X M+

Xx MM' Xx MM

M+ X M+ X

M+ X M+ X

MM Xx MM Xx

X M+ X M+

X M+ Xo M+

Xx MM Xx MM

|M|'
M

e'

FIGURE 8.2 Perfect (left) and defective crystal situations for the compound M+X. A specific example may be Ag+Cl. Top row: ionic defects.
Bottom row: electronic defects. First and second columns: structure elements in absolute notation. Th ird column: structure elements in relative
notation. Fourth column: building elements.

8-3

Nanoionics
Emax
E

Excited states

Problem

Emin

Level distr.

Eon, deloc.

Ion, crystal

= E+ kT

()

Ion, amorphous

Ground states

G= EN kT ln

()

Z
N

Emax

N=

N/Z

Emin

N
Z

Z/E

Parabolic

m3/2

Delta

Gaussian

Emin

Z
dE~
N exp
E
kT

= Emin + kT ln(N/N )

1 N/Z

FIGURE 8.3 Whenever the gap between ground states and excited states is large, a Boltzmann distribution results. Details on the density of states
enter the constant term (also the concentration measure). Eon stands for electron. If the reader is interested in more details, they are referred to
Maier (2005a) and Kirchheim (1988).

Local partial free energy

Interstitital ionic level ~


i
Regular ionic level

~
M+

0
Valence band

~
e
Conduction band

(a)

Rule of
homogeneous doping

~
n

zkck
zC

<0

ck (P ,T , C) = k k P N k C Mk r K r rk (T ),

(b)

FIGURE 8.4 (a) Defect thermodynamics of a mixed conductor


with Frenkel disorder (Maier 2003a). (b) Rule of homogeneous doping. (Reprinted from Maier, J., Solid State Ionics, 157, 327, 2003a. With
permission.)

concepts.) Hence the distances between Frenkel (Fermi) levels and


energy levels (band edges) are measures of interstitial and vacancy
(hole and electron) concentrations.
Figure 8.4 also displays the coupling between both pictures
that is provided by the relation

+
= M+ + e = M ,
M+
e

determined through M (=const + RT ln PM) and hence by the


partial pressure of M (i.e., PM). In a binary oxide, it is convenient
and sufficient to refer to PO2 (that is coupled to PM via the phase
equilibrium conditions).
Further important parameters for the equilibrium defect
chemistry are temperature and doping content (C). Our energy
level picture is only able to give an account of simple situations.
A complex set of processes is better handled by the language of
chemical thermodynamics, i.e., by explicitly writing down interaction equilibria to which then mass action laws apply. In not
too complex cases, it is easy to show that the concentration for a
charge carrier k, i.e., ck(P, T, C), is given by

(8.1)

where and ~
are the chemical and electrochemical potentials
(the latter include electrical potentials).
As M = G / nM measures the increase of Gibbs energy
if neutral M is added to the master compound MX (n = mole
number) at constant nX and hence refers to compositional variations, this shows how important stoichiometry is (adding M to
MX leads to variations in the M/X ratio).The stoichiometry is

(8.2)

where
, , N, M, are simple rational numbers
Kr(T) is the mass action constant of the defect reaction r
In the simplest case, see Figures 8.3 and 8.4, we only have to
refer to a single r as in the case of our energy level diagram. Th is
equation looks slightly more complicated if more components
are involved. If these additional components, i are also in equilibrium with k then P N has to be replaced by a product of terms
( i Pi Ni ). If the other components (j with charge number zj) are
frozen however, then the respective defect concentrations (cj)
appear in the C-term as dopants (jzjcj). Let us briefly consider
the influence of these parameters on a qualitative level (see Maier
(2004b) for a quantitative treatment):
1. Temperature increase typically (but not in all cases) leads
to higher defect concentrations as the formation processes
are all thermally activated. Of special significance here
are high defect concentrations, where interactions occur.

8-4

Handbook of Nanophysics: Principles and Methods

Perfect

Weakly defective

Heavily defective

T=0 K

0 < T << Tc

T<
~ Tc

Heavily defective

Weakly defective
~
E =
Excited

Perfect

Interstitial

~
Ereg =
v

Regular

~
i + RT ln yei

~ RT ln ye

v
v

T > Tc

Superionic

Ei(c)

~ Ereg
Ei

Ereg(c)

T<
~ Tc

0 < T << Tc

Superionic

T > Tc

FIGURE 8.5 Thermal energy leads to Frenkel disorder. Further increase leads to attractive interactions. The positive feedback leads to
super-ionic phase transformation. (Bottom: energy level description). (Reprinted from Maier, J. and Mnch, W., Z. Anorg. Allg. Chem., 626, 264,
2000. With permission.)

Interactions between intrinsic majority effects are usually


attractive and can lead to the transition of the superionic
state, in which all ions are defective, as it were. This is
shown in Figure 8.5. In this case, however, we are dealing
already with deviations from Boltzmann distribution and
from Equation 8.2.
2. Figure 8.6 shows the drastic variation in ionic and electronic charge carrier concentration upon tiny changes in
the partial pressure.
3. Figure 8.7 illustrates the effect of doping. An effectively
negatively charged dopant leads to increase of all the positively charged defects (here VO , h ) and to a decrease of
all negative charged defects (here Oi, e) individually. This
can be summarized by the rule of homogeneous doping
(already indicated in Figure 8.4)

(8.3)

The presence of point defects is of course not sufficient for


having a perceptible conductivity (). For this, we need also a
sufficient mobility (u) as k ukck.
Figure 8.8 displays the three basic ionic jump mechanisms:
the transport of a vacancy by allowing neighboring regular
ions to occupy it (top), and the transport of interstitials that
either jump directly to the next equivalent site (center) or
indirectly by substituting a regular neighbor that is to occupy
another interstitial site (bottom). Obviously, the respective
ionic mobilities are all strongly thermally activated because of
the pronounced migration barrier. As stoichiometric or dopant

log (|z| [defectz])

(1/6)

Oi''

O''i

(0)

V O, Oi''

(1/6)

(1/4)
e'

VO

P(i)
log PO2

z k ck
< 0.
zC

VO
e'

= [Oi''] [VO]
=

MO

[h ] [e']
2

PO

FIGURE 8.6 Defect chemical variation in the oxygen superlattice disordered oxide MO by PO2 variation. (Reprinted from Maier, J.,
Modern Aspects of Electrochemistry, Conway, B.E. et al. (eds.), Vol. 38,
Kluwer Academic/Plenum Publishers, New York, 2005c, 1173. With
permission.)

8-5

Nanoionics

A', VO

log |z| [defect]

1
VO , Oi''

8.3 Ionic Charge Carrier Distribution at


Interfaces and Conductivity Effects

+1/2

Oi''

e'

1/2

A'
log [A']

FIGURE 8.7 Acceptor doping of an oxide with oxygen sublattice disorder leads to an increase of oxygen vacancy and hole concentrations
(while the concentrations of interstitial oxygen ions and electrons are
depressed).

(a)

As conductivity is proportional to the product of mobility and


concentration, conductivity variations at the interfaces can be
due to both factors. Directly in the interfacial core, these two
parameters are naturally very different from the bulk. This influence is particularly decisive if we consider materials with low
charge carrier conductivities in the bulk. A significant effect
whose importance can be hardly overestimated is the charge
carrier redistribution in the vicinity of the interfaces. Naturally,
one may also meet variations of mobility due to structural variations that are expected to be minor compared to the interfacial
core structure (see below). Such mobility variations are particularly important if the charge carrier concentration is already
very high and bulk mobilities low, while concentration changes
particularly matter for materials with high carrier mobilities
and low bulk concentrations. Under these conditions, space
charge effects may be order of magnitude effects. We will largely
dwell on such space charge effects (Maier 1995) owing to their
importance and generality of the phenomenon.
For a phenomenological description, let us consider (see
Figure 8.9) the general thermodynamic situation in a mixed conductor at the interface in the approximation of an abrupt contact
(i.e. we assume that the structure varies in a step-function way at
x = 0 toward the bulk structure).
The electronic levels have to be bent and so do the levels of the
mobile ions. This is equivalent to having concentration changes
of both ions and electrons (Figure 8.10). If we introduce conductivity effects owing to the introduction of appropriate second
phases, we term this heterogeneous doping. The rule of heterogeneous doping, which proposes how a given surface charge
() qualitatively affects the concentration of a given defect k is
analogous to Equation 8.3 (see also Figure 8.9), namely,

(b)

(c)

FIGURE 8.8 Three basic ion transport mechanism (a) vacancy,


(b) direct interstitial, and (c) indirect interstitial. (Reprinted from
Maier, J., Physical Chemistry of Ionic Materials. Ions and Electrons
in Solids, John Wiley & Sons, Ltd., Chichester, U.K., 2004b. With
permission.)

variations affect the structure only at high concentrations, the


mobility is taken as independent of P and C in the following if
not stated otherwise.
With these points in mind, the situation at interfaces can be
tackled.

Local partial free energy

Interstitial ionic level

Regular ionic level

~
v

~
M+

0
Valence band

~
e
Conduction band

(a)

~
i

Rule of
heterogeneous doping

~
n

zkkck

<0

(b)

FIGURE 8.9 (a) General contact of a Frenkel disordered mixed conductor to a second phase, resulting in a negative surface charge. (b) Rule
of heterogeneous doping. (Reprinted from Maier, J., Modern Aspects of
Electrochemistry, Conway, B.E. et al. (eds.), Vol. 38, Kluwer Academic/
Plenum Publishers, New York, 2005c, 1173. With permission.)

8-6

VAg

15

1, 0

VO

AgCl

Exp. results and


theor. calculations

SrTiO3

0
A

A
Vg

VO
e

AgCl
0

CeO2
0

Distance from the interface

Distance from the interface

lg (m/1 cm1)

Agi

10

4
0

Exp. results and


theor. calculations

u
uv
m = (1 ) + L F (2)c 2 v v

+
v
1
1
v

where
is the Debye length
is the degree of influence
L is the percolativity

(8.5)

(8.4)

3 T 1/K1

10

FIGURE 8.11 Heterogeneous doping of AgCl and AgBr with Al 2O3


and detailed description by Equation 8.5 (according to Equation 8.5
1). (Reprinted from Maier, J., Prog. Solid State Chem., 23, 171, 1995.
With permission.)

5 103
Zero bias
200 mV
400 mV
600 mV

T = 598 K
PO2 = 105 Pa

4 103

lm Z /

Figure 8.10 shows some realistic examples for a positively charged


surface. On a qualitative level, all these results can be understood on the basis of Equation 8.4. Figure 8.10 (top, left) refers to
an accumulation of vacancies and an increased ion conductivity
in a donor doped sample (example: AgCl:Al2O3 or AgCl:AgCl,
a typical donor being Cd2+) (Maier 1995), Figure 8.10 (top, right)
refers to a depletion of both vacancies and holes as met in acceptor doped SrTiO3 ceramics (a typical acceptor is Fe3+) (Vollmann
and Waser 1994; Noll et al. 1996; Vollmann et al. 1997; De Souza
et al. 2003). Also inversion effects can be realized, e.g., in acceptor (S2)-doped AgCl, where in the bulk we meet predominant
interstitial conductivity but at the boundary vacancy conductivity, Figure 8.10 (bottom, left). In Figure 8.10 (bottom, right) we
face even an inversion from ion (vacancy) to electron (n-type)
conductivity in weakly acceptor (Gd)-doped ceria (Chiang et al.
1996; Tschpe 2001; Kim and Maier 2002).
Figures 8.11 and 8.12 give experimental examples. Figure 8.11
shows a significant enhancement that is achieved by adding Al2O3
particles (with volume fraction and surface-to-volume ratio )
to AgCl, which leads to an adsorption of Ag+ ions at the aluminas
surface connected with an accumulation of carriers. In percolative
composites, the conductivity can be derived to be (Maier 1995)

10
7
(13)

AgCl: Al2O3

z k ck
< 0.

15
10 6
5
(10)
1

FIGURE 8.10 Four boundary situations at a positively charged contact (see text). The influence of charge number (+2 for oxygen vacancies,
1 for e, h, Ag+-defects) is seen in the bulk balance as well as in the
steepness of variation at the boundary.

lg (m/1 cm1)

AgBr: Al2O3

Log defect concentration

Log defect concentration

Handbook of Nanophysics: Principles and Methods

3 103
2 103

20 Hz

1 MHz

1 103

1 103

3 103

5 103

7 103

Re Z /

FIGURE 8.12 Bias-dependent impedance spectra of a SrTiO3 bicrystal. The bicircles on the right reflect hole depletion. (Reprinted from
Denk, I. et al., J. Electrochem. Soc., 144, 3526, 1997. With permission.)

Figure 8.11 shows how accurately this equation describes


the experimental results. If the AgCl is pure or S2-doped, the
bulk conductivity is of interstitial type; then we have realized an
interstitial-vacancy junction by heterogeneous doping. If we consider TlCl, which intrinsically conducts via Cl vacancies, then
heterogeneous doping by Al2O3 leads to a predominant cationic

8-7

Nanoionics
0

~9 nm
2
log (T/S cm1 K)

Mass stored

Conductance

B
5

Volume fraction of B

FIGURE 8.13 Conductance and mass storage anomaly in an A:B twophase mixture owing to ion redistribution. (Reprinted from Zhukovskii,
Y.F. et al., Phys. Rev. Lett., 96, 058302, 2006. With permission.)

7
1.0

1.2

1.4

1.6

1.8

2.0

2.2

2.4

103 T 1/K1

FIGURE 8.14 Fluoride conductivity of nano- and macrocrystalline


CaF2. (Reprinted from Puin, W. et al., Solid State Ionics, 131, 159, 2000.
With permission.)

2.40E009

Normalized conductance Y ||/1

conductivity (via cation vacancies) (Maier 1995). This effect of


heterogeneous doping is qualitatively sketched in Figure 8.13.
The depletion effect on the conductivity in the primarily
electronically conducting (acceptor-doped) SrTiO3 is directly
reflected by the low frequency semicircle of a bicrystal impedance experiment (Figure 8.12) (Noll et al. 1996; Merkle and Maier
2008). The fact that also and even more strongly oxygen vacancies ( VO ) are depleted is shown by measurements in which the
electronic conductivity has been blocked. (The stronger depletion of VO as compared with h leads to peculiar polarization
phenomena in multilayers or polycrystalline material under dc
bias (Jamnik et al. 2003).)
An even more striking situation is met for a vacancy conducting
oxide with excess electrons as major electronic carriers. In CeO2
(weakly doped by Gd2O3) clearly a transition from ionic to electronic conductivity occurs on introducing appropriate interfaces.
A key point is the mechanism that determines the space charge
potential. This obviously lies in the chemistry/crystallography of
the contact (Maier 1995). Understanding this in principle would
open the possibility of tuning the space charge potential. For a
quantitative description, the reader is referred to the literature.
One example in which a purposeful tuning succeeded refers to
polycrystalline or thin fi lms of CaF2, which is a fluoride ion conductor. At interfaces, fluoride vacancy accumulation has been
observed that could purposefully be augmented by contaminating the interfaces with an F attractor (SbF5 or BF3) (Saito
and Maier 1995). Figure 8.14 shows the enormous conductivity
enhancement of nanocrystalline CaF2 (Puin et al. 2000), where it
is the enormous proportion of interfaces that leads to these huge
effects. Figure 8.15 gives results on thin fi lms of BaF2 on Al2O3substrate, which leads to a fluoride vacancy ( V F ) accumulation
(Guo and Maier 2004).
In all the examples considered, the quantitative explanation
relied on an interfacial situation in which the carrier profi le
reached bulk values; in other words, the local situation was semiinfinite. In the following, we will treat even more exciting effects
resulting from interfacial overlap. Before we do so, let us mention

m-CaF2 (dgrain = 200 nm)


n-CaF2 (dgrain = 9 nm)
Calculated for one monolayer at n-CaF2-interfaces

1.60E009

8.00E010

T = 510C
BaF2/Al2O3(012)
(a)

0.00E+000
8.00E011

4.00E011
T = 370C
BaF2/Al2O3(012)
(b)
0.00E+000

100

200
Thickness/nm

300

FIGURE 8.15 Parallel conductance of BaF2 fi lms on Al 2O3 substrates.


The explanation is based on F-adsorption. At 370C strain effects contribution. (Reprinted from Guo, X.X. and Maier, J., Surf. Sci., 549, 211,
2004. With permission.)

one more point that is interesting for semiconductor physics and


follows from the fact that majority defects typically control the
space charge behavior. Also in typical semiconductors, we meet
ionic point defects, very often in much greater concentrations
than electrons. Owing to their low mobility, in particular at the
low temperatures, they do not actually contribute to the conduction process, but act as dopants. Their concentration is typically

8-8

Handbook of Nanophysics: Principles and Methods

determined by the prehistory. This is very relevant for the boundary effects in particular when pretreatment at high temperatures
was performed at which the ions are locally mobile. Then very
often, the ionic boundary behavior determines the space charge
potential that also has to be obeyed by the electrons as minority species. This fellow traveler effect (Maier 1995) can explain
space charge effects of electronic carriers that are not compatible with or even seemingly in contradiction with the expected
polarity. It is not hopeless to use this effect for tuning in future
high-temperature superconduction.
Before we deal with mesoscopic phenomena, let us consider
Table 8.1, which gives results for the conductivity effect of ions
and/or electrons depending on the direction of measurements
(Kim et al. 2003). It is clear that in a measurement direction
along the interface, the highly conductive regions are seen most
sensitively, whereas it is the most resistive regions that are dominant in a measurement perpendicular to the interface.

8.4 Mesoscopic Effects


1. A well-suited model system that allows studying nanoionic effects in detail are molecular beam epitaxially
grown heterolayers consisting of CaF2 and BaF2 (Sata et al.
2000; Guo et al. 2007; Guo and Maier 2009). The thickness of the individual layers can be varied from almost
1 nm to 1 m. The misfit stress is absorbed by misfit
dislocations. It was shown that the conductivity effects
are purely ionic. Impedance spectroscopy is applied to
measure parallel as well as perpendicular conductivities
using appropriate substrates. All the effects can be consistently explained by a fluoride ion transfer from BaF2 to
CaF2, hence increasing the vacancy concentration in BaF2
and the interstitial concentration in CaF2. A fully quantitative explanation of thickness and temperature variation
dependences succeeds if one takes account of background

TABLE 8.1 Effective Conductivities and Resistivities for Experiments Parallel and
Perpendicular to an Interface of a Mixed Conductor (Example: Weakly Acceptor Doped CeO2
and Positively Charged Interface; Here the Parallel Conductivity is Dominated by Electrons
(n) and the Perpendicular Resistivity by the Vacancies (v))

Model

||
Effective Parallel Conductivity, m
and
Perpendicular Resistivity, m

Concentration Profile
A

SchottkyMott

||

e
VO

cn0

m,n

m,v

||
m,n

(2) cn0 cn

m,v

(2)

m,n

||

(2)

m,v

(2)

||
m,n

(2) cn0 cv

m,v

(2)

2 ln(cn0/cn)
1
2cv0 ln(cv0/cv)

*
GouyChapman

VO

1
cv0 cv

cn0 cn

VO

1
cv0 cv

cn0
1
cv0 cv

1
cv

cn0 cn

1
cv0 cv

Combined

VO

1
cv0 cn

cn0
1
cv0

1
cv cn

Source: Reprinted from Kim, S. et al., Phys. Chem. Chem. Phys., 5, 2268, 2003. With permission.

8-9

Nanoionics
16,000
T = 593 K
(320C)

120

T = 773 K
(500C)

m, total||

|| /106 1 cm1
m

|| /106 1 cm1
m

12,000
80

40

||

m, BaF2
8,000

4,000

||

m, CaF2
0

0
0

1 106

2 106

3 106

4 106

Inverse interfacial spacing/cm

Inverse interfacial spacing/cm

8 106
1

FIGURE 8.16 Parallel conductivity of CaF2 and BaF2 multilayers as a function of inverse spacing. The characteristic varies from MottSchottky
type to GouyChapman type if temperature is increased from 320C to 500C. (Reprinted from Guo, X.X. and Maier, J., Adv. Funct. Mater., 19, 96,
2009. With permission.)

a synergetic storage is possible that can be appreciable in


amount and speed. In the mesoscopic regime, this jobsharing provides the bridge between an electrostatic
capacitance and a battery capacitance (Zhukovskii et al.
2006).
A particularly well-suited material in this respect is a
nanocomposite of Li2O and Ru. Owing to selective storage
ability of Li+ in Li2O and e in Ru, nanocomposites can
store a lot of Li with a pseudocapacitive charge/discharge
behavior. The storage is also quick as Li+ will take the Li2O
route and the electron metal route. Figure 8.17 gives the
chemical potential for such a storage situation, showing
that unlike bulk storage, Li storage is in this case not connected with a local variation in Li (Maier 2007a).
M

Li2O

Li
Chemical potential, j

impurities. The treatment is even able to explain annealing effects during preparation. Let us focus on the effective
parallel conductivity and its dependence on the density

of the interfaces ((thickness)1). Initially m varies linearly with L1 while the nonlinear increase can be nicely
connected with space charge overlap (Figure 8.16). (The
upward bending occurs in a MottSchottky situation but
disappears at high temperature where intrinsic disorder
overwhelms and a GouyChapman layer is established.)
There are also a variety of studies of oxide heterolayers,
the most careful of which were performed by Korte et al.
(2008, 2009). They use Y-ZrO2, which due to its high disorder is not expected to show an accumulation layer that
could give rise to great concentration enhancement. Here
it is anticipated that mobility effects are observable. Indeed
Korte et al. could show that the moderate interfacial effect
scales with the degree of misorientation. Tensile strain
increases the mobility of ZrO2, while compressive strain
decreases it in agreement with the migration-volume of
vacancy transport. The samples investigated are not in the
regime of interfacial overlap. Mesoscopic strain effects are
involved in the low temperature example of Figure 8.15.
2. Overlap of depletion zones is detected in nanocrystalline
ceramics of SrTiO3. The bulk semicircle still seen in Figure
8.12 now has disappeared. More detailed modeling shows
that this is not a matter of resolution rather the depletion
zone extends throughout. Interestingly, the respective
capacitance that was the bias-dependent interfacial capacitance for macrocrystalline SrTiO3 is now bias independent as the surface charge becomes invariant (Balaya et al.
2006).
3. Space charge effects also can be relevant for storage: a
composite of and can store a compound A+B even if
none of the phases and can do this individually (Figure
8.13). If can store A+ (not B) and can store B (not A+)

Li+

Li+
Position coordinate

FIGURE 8.17 Excess storage isas it occurs heterogeneouslynot


connected with a variation in the chemical potential of lithium (Li).
(Reprinted from Maier, J., Faraday Discuss., 134, 51, 2007a. With
permission.)

8-10

Handbook of Nanophysics: Principles and Methods

Distance from interface

FIGURE 8.18 Confi nement of localized point defects and of delocalized electrons (see text). (Reprinted from Maier, J., Modern Aspects of
Electrochemistry, Conway, B.E. et al. (eds.), Vol. 38, Kluwer Academic/Plenum Publishers, New York, 2005c, 1173. With permission.)

4. So far, we assumed the energy level (i.e., the local partial


free energy) of the point defect to be undisturbed. Clearly,
a variation will occur at tiny sizes (Figure 8.18). The analogous effect for electrons is well known for the electrons in
the context of the particle-in-the-box-problem.
5. Here we have to essentially deal with classical electrostatic
polarization effects that can be of longer range than the
dimension of the sheer point defect. If this perturbation
zone is of the order of the confinement, we expect a variation of (Maier 2003b, 2004a, 2005a). As electrons and
ions perceive confinement differently, the bending need
not be parallel any longer.
6. So far we excluded structural variations other than the
abrupt junction. If elastic effects lead to structural variations inside the phase under consideration, this is naturally
also perceived by the defects to be formed within this
ground structure.
7. Another effect that is known from semiconductor physics is that electrons and holes cannot evade each other in
sufficiently confined systems. Then, owing to Coulomb
attractions, the effective band gap decreases. The same
phenomenon is expected for interstitials and vacancies
in ion conductors. As such interactions are connected
with phase transitions to the superionic state, also transition temperatures may be affected (Maier 2003b, 2004a,
2005a).
8. Furthermore configurational effects are expected at minute size. As this is more important for tiny crystallites, we
will come back to this point in the following section where
we take explicit account of curvature effects.

8.5 Consequences of Curvature


for Nanoionics
Unlike fi lms or macroparticles, nanoparticles are heavily curved
(radius: r) and therefore exhibit a perceptibly increased internal
(capillary) pressure. Ignoring stress effects, the chemical potential is increased by (Defay et al. 1960)
k (r ) = k () + 2Vk .

(8.6)

As now no characteristic layers can be defined like the Debye


length for space charges, the curvature influence escapes the distinction between trivial and nontrivial size effects.
If stress effects are negligible and morphological equilibrium
is fully established, is the Wulff ratio i / ri (i: surface tension
of surface plane with orientation i, r: the length of the normal
vector pointing from the centre of the crystal to the surface
plane), which is for a Wulff-shaped crystal the same for any
orientation i and can also be replaced by area averaged quantities, i.e., by / r . Owing to the coherency between surface and
bulk, stress effects are, however, important. Then the surface
stress tensor becomes a pertinent capillary parameter (Rusanov
1995; Kramer and Weissmller 2007). Moreover, deviations
from the Wulff shape necessarily occur and the interpretation
of becomes complicated. At any rate, capillary effects lead to
drastic consequences such as depression of the melting point of
a small nanoparticle. Capillarity also leads to a nonzero open
circuit voltage for the symmetrical cell (Schroeder et al. 2004):
Ag(nano)|Ag + -conductor|Ag(macro).
A capillarity term also varies the chemical potential of a defect
(Vk then being the partial molar volume of the defect), which is
expected to lead (i) to varied defect concentrations at the same control parameters, (ii) to a charging at the contact of two differently
curved identical particles, in particular if a small nanocrystal is
in contact with a film of the same chemistry (Figure 8.19) (Maier
2002, 2003a,b, 2004a).
There is an important difference in the stability of nanoparticles depending on the fact whether elementary normal or
binary (ternary etc.) particles are concerned. If M particles that
do not undergo normal Ostwald ripening are contacted with an
M+ electrolyte, electrochemical Ostwald ripening can take place
for kinetic reasons (Schrder et al. 2006). This is different if additional immobile elements constitute the compounds.
Finally it should be mentioned that phase transition characteristics look different on the nanoscale: (i) If in a two-phase regime
the overall composition is varied, in macrocrystalline cases, the
chemical potential of an exchangeable compound stays constant,
as only the relative proportion of the two coexisting phases is

8-11

Nanoionics

M+

MX
MX

FIGURE 8.19 Capillary effects demand a space charge effect just


because of different curvature (Maier 2003b). (Reprinted from Maier, J.,
Z. Phys. Chem., 219, 35, 2005a. With permission.)

changed but not concentrations therein. This may be different


on the nanoscale as sizes may change. (ii) Let us consider a first
order transition (at Tc) in a macroscopic system. Below Tc the
total solid is in the lower energy state, while above Tc the solid
is in the upper level. This is different in molecular systems, in
which always a fraction even below Tc may occupy the higher
level; this is because the configurational entropy is not negligible
compared to the energy difference. Nanocrystals if sufficiently
small should behave as intermediates. It has however to be borne
in mind that the configurational entropy rapidly decreases with
increasing size (Hill 1963; Maier 2009).
How statistics have to be performed over a nanosized ensemble is generally a fascinating but not fully explored subject. All
the points addressed become very different if we leave the regime
of solid state physics and chemistry where each size requires a
thermodynamically different equilibrium structure. Then finally
we have entered the regime of cluster physics and chemistry and
left our topic nanoionics.

8.6 Conclusions
Various examples have been considered that showed the significance of small size for ion transport and mass storage properties. A variety of basic scientific problems are connected with
this. Though the area of nanoionics is just in its infancy,
already the recent research did not only reveal conductivity and
storage phenomena that are order-of-magnitude effects, but
also a variety of implications have been identified that are of
direct importance for applications, in particular for Li-battery
technology (Maier 2007b).

References
Balaya, P., Jamnik, J., Fleig, J., and Maier, J. 2006. Mesoscopic
electrical conduction in nanocrystalline SrTiO3. Appl. Phys.
Lett. 88: 062109.
Chiang, Y. M., Lavik, E. B., Kosacki, I., Tuller, H. L., and Ying, J. Y.
1996. Defect and transport properties of nanocrystalline
CeO2-x. Appl. Phys. Lett. 69: 185187.

De Souza, R. A., Fleig, J., Maier, J., Kienzle, O., Zhang, Z., Sigle, W.,
and Rhle, R. 2003. Electrical and structural characterization of a low-angle tilt grain boundary in iron-doped strontium titanate. J. Am. Ceram. Soc. 86: 922928.
Defay, R., Prigogine, I., Bellemans, A., and Everett, H. 1960.
Surface Tension and Adsorption. New York: John Wiley &
Sons, Ltd.
Denk, I., Claus, J., and Maier, J. 1997. Electrochemical investigations of SrTiO3 boundaries. J. Electrochem. Soc. 144:
35263536.
Guo, X. X. and Maier, J. 2004. Ionic conductivity of epitactic
MBE-grown BaF2 films. Surf. Sci. 549: 211216.
Guo, X. X. and Maier, J. 2009. Comprehensive modeling of ion
conduction of nanosized CaF2/BaF2 multilayer heterostructures. Adv. Funct. Mater. 19: 96101.
Guo, X. X., Matei, I., Jamnik, J., Lee, J. S., and Maier, J. 2007.
Defect chemical modelling of mesoscopic ion conduction
in nanosized CaF2/BaF2 multilayer heterostructures. Phys.
Rev. B 76: 125429 (17).
Hill, L. 1963. Thermodynamics of Small Systems. New York:
W. A. Benjamin Inc. Publ.
Jamnik, J., Guo, X., and Maier, J. 2003. Field-induced relaxation
of bulk composition due to internal boundaries. Appl. Phys.
Lett. 82: 28202822.
Kim, S. and Maier, J. 2002. On the conductivity mechanism of
nanocrystalline ceria. J. Electrochem. Soc. 149: J73J83.
Kim, S., Fleig, J., and Maier, J. 2003. Space charge conduction:
Simple analytical solutions for ionic and mixed conductors
and application to nanocrystalline ceria. Phys. Chem. Chem.
Phys. 5: 22682273.
Kirchheim, R. 1988. Hydrogen solubility and diffusivity in defective and amorphous metals. Prog. Mater. Sci. 32: 261325.
Korte, C., Peters, A., Janek, J., Hesse, D., and Zakharov, N. 2008.
Ionic conductivity and activation energy for oxygen ion
transport in superlatticesthe semicoherent multilayer
system YSZ (ZrO2 + 9.5 mol% Y2O3)/Y2O3. Phys. Chem.
Chem. Phys. 10: 46234635.
Korte, C., Schichte, N., Hesse, D., and Janek, J. 2009. Influence of
interface structure on mass transport in phase boundaries
between different ionic materials. Experimental studies and
formal considerations. Chem. Monthly 140: 10681080.
Kramer, D. and Weissmller, J. 2007. A note on surface stress and
surface tension and their interrelation via Shuttleworths
equation and the Lippmann equation. Surf. Sci. 601:
30423051.
Maier, J. 1995. Ionic conduction in space charge regions. Prog.
Solid State Chem. 23: 171263.
Maier, J. 2002. Thermodynamic aspects and morphology of nanostructured ion conductors. Aspects of nano-ionics. Part I.
Solid State Ionics 154155: 291301.
Maier, J. 2003a. Defect chemistry and ion transport in nanostructured materials. Part II. Aspects of nanoionics. Solid
State Ionics 157: 327334.
Maier, J. 2003b. Nano-Ionics: Trivial and non-trivial size effects
on ion conduction in solids. Z. Phys. Chem. 217: 415436.

8-12

Maier, J. 2004a. Ionic transport in nano-sized systems. Solid State


Ionics 175: 712.
Maier, J. 2004b. Physical Chemistry of Ionic Materials. Ions and
Electrons in Solids. Chichester, U.K.: John Wiley & Sons, Ltd.
Maier, J. 2005a. Chemical potential of charge carriers in solids.
Z. Phys. Chem. 219: 3546.
Maier, J. 2005b. Nanoionics: Ion transport and electrochemical
storage in confined systems. Nat. Mater. 4: 805815.
Maier, J. 2005c. Solid state electrochemistry I: Thermodynamics
and kinetics of charge carriers in solids. In Modern Aspects
of Electrochemistry, B. E. Conway, C. G. Vayenas, and
R. E. White (Eds.), vol. 38, pp. 1173. New York: Kluwer
Academic/Plenum Publishers.
Maier, J. 2007a. Mass storage in space charge regions of nanosized systems (Nano-ionics Part V). Faraday Discuss. 134:
5156.
Maier, J. 2007b. Size effects on mass transport and storage in lithium batteries. J. Power Sources 174: 569574.
Maier, J. 2009. Thermodynamics of nanosystems with a special
view to charge carriers. Adv. Mater. 21: 25712585.
Maier, J. and Mnch, W. 2000. Thermal destiny of an ionic crystal.
Z. Anorg. Allg. Chem. 626: 264269.
Merkle, R. and Maier, J. 2008. How is oxygen incorporated into
oxides? A comprehensive kinetic study of a simple solidstate reaction with SrTiO3 as a model material. Angew.
Chem. Int. Ed. Engl. 47: 38743894.
Noll, F., Mnch, W., Denk, I., and Maier, J. 1996. SrTiO3 as a prototype of a mixed conductor: Conductivities, oxygen diffusion,
and boundary effects. Solid State Ionics 8688: 711717.
Puin, W., Rodewald, S., Ramlau, R., Heitjans, P., and Maier, J.
2000. Local and overall ionic conductivity in nanocrystalline CaF2. Solid State Ionics 131: 159164.
Rusanov, A. I. 1995. Interfacial thermodynamics: Development
for last decades. Solid State Ionics 75: 275279.

Handbook of Nanophysics: Principles and Methods

Saito, Y. and Maier, J. 1995. Ionic conductivity enhancement of


the fluoride conductor CaF2 by grain boundary activation
using Lewis acids. J. Electrochem. Soc. 142: 30783083.
Sata, N., Eberman, K., Eberl, K., and Maier, J. 2000. Mesoscopic
fast ion conduction in nanometre-scale planar heterostructures. Nature 408: 946949.
Schroeder, A., Fleig, J., Drings, H., Wuerschum, R., Maier, J., and
Sitte, W. 2004. Excess free enthalpy of nanocrystalline silver,
determined in a solid electrolyte cell. Solid State Ionics 173:
95101.
Schrder, A., Fleig, J., Gryaznov, D., Maier, J., and Sitte, W. 2006.
Quantitative model of electrochemical Ostwald ripening and its application to the time-dependent electrode
potential of nanocrystalline metals. J. Phys. Chem. B 110:
1227412280.
Tschpe, A. 2001. Grain size-dependent electrical conductivity of
polycrystalline cerium oxide II: Space charge model. Solid
State Ionics 139: 267280.
Vollmann, M. and Waser, R. 1994. Grain boundary defect chemistry of acceptor-doped titanates: Space charge layer width.
J. Am. Ceram. Soc. 77: 235243.
Vollmann, M., Hagenbeck, R., and Waser, R. 1997. Grainboundary defect chemistry of acceptor-doped titanates:
Inversion layer and low-field conduction. J. Am. Ceram.
Soc. 80: 23012314.
Wagner, C. and Schottky, W. 1930. Theorie der geordneten
Mischphasen. Z. Phys. Chem. B 11: 163210.
Zhukovskii, Y. F., Balaya, P., Kotomin, E. A., and Maier, J. 2006.
Evidence for interfacial-storage anomaly in nanocomposites for lithium batteries from first-principles simulations.
Phys. Rev. Lett. 96: 058302 (14).

You might also like