This journal is cthe Owner Societies 2011 Phys. Chem. Chem. Phys., 2011, 13, 1239–1242 1239
An EMF cell with a nitrogen solid electrolyte—on the transference
of nitrogen ions in yttria-stabilized zirconia
Doh-Kwon Lee,w
a
Claus C. Fischer,
a
Ilia Valov,z
a
Jochen Reinacher,
a
Alexandra Stork,
b
Martin Lerch
b
and Juergen Janek*
a
Received 9th March 2010, Accepted 14th October 2010
DOI: 10.1039/c003991h
The mobility and electrochemical activity of nitrogen inside and/or at the surface of ionic
compounds is of fundamental, as well as of possibly practical, relevance. In order to better
understand the role of nitrogen anions in solid electrolytes, we measured the transference number
of nitrogen in yttria-stabilized zirconia (YSZ) by a concentration cell technique as a function
of oxygen activity at different temperatures in the range of 1023 rT/Kr1123. YSZ doped with
1.9 wt% of N (YSZ:N) turned out to have an appreciable nitrogen transference number, which
increased from 0 to 0.1 with decreasing oxygen activity in the range of 20 ologa
O
2
o14.
The stability of N in YSZ:N, however, has yet to be elucidated under oxidizing conditions.
1. Introduction
Nitrogen is usually regarded as (electro)chemically inert at
most gas/solid interfaces or electrodes. This is partly because
there has been no nitrogen solid electrolyte available so far.
Only recently was nitrogen doping into cubic stabilized zirconia
attempted as a possible alternative to cation doping, in light of
the increase in oxygen vacancy concentration as well as phase
stabilization.
1–11
Whether the resulting crystalline and homo-
geneous quaternary fluorite-type oxide nitride phases can
indeed be used as nitrogen electrolytes remains an open
question to date.
Yttria-stabilized zirconia (YSZ) is one of the most intensively
studied oxide ion conductors with an oxygen transference
number t
O
2
(s
O
2
/s
tot
,wheres
O
2
and s
tot
stand for the
partial conductivity of oxygen ions and the total conductivity,
respectively) close to unity. Hence, it is used as a solid electrolyte
in a variety of applications, such as solid oxide fuel cells (SOFC),
oxygen sensors, pumps or membranes, to name a few. The
electrolytic functionality is facilitated by a large number of mobile
oxygen vacancies formed to maintain the charge-neutrality with
Y
3+
ions in the Zr
4+
sublattice. The concentration of oxygen
vacancies can be additionally increased by heterogeneous nitrogen
incorporation, which essentially leads to the substitution of
oxygen, as represented in Kroeger–Vink notation
1,5
N2ðgÞþ3O
O¼2N0OþV
Oþ3
2O2ðgÞ;ð1Þ
with N0Odenoting the N
3
ion on an O
2
site, V
Odenoting
the oxygen vacancy and O
Orepresenting the O
2
ion on a
regular anion site. Accordingly, N-doped YSZ has been found
to have a higher ionic conductivity than ‘‘undoped’’, i.e.
nitrogen-free, YSZ at high temperatures above 1000 1C.
5
At
lower temperatures, the ionic conductivity is poorer, probably
due to strong interactions between nitrogen ions and
vacancies. Up to now, the transport properties of YSZ:N were
only reported in terms of total conductivity. Little is known
about the mobility of nitrogen itself.
6,12,13
In this paper, we
report on the non-zero mobility of nitrogen in N-doped YSZ,
determined by an open cell voltage measurement as a function
of oxygen activity and temperature. The results also support
our earlier observation that dinitrogen is electrochemically
active at metal(N
2
)/solid electrolyte electrodes.
14,15
2. Principles
The open cell voltage measurement based on a galvanic cell in
a chemical potential gradient (concentration cell) is usually
employed to determine the transference number of ionic
species in ionic compounds. When an oxide is placed under
a chemical potential difference of a thermodynamic com-
ponent, an electromotive force E(emf, open cell voltage or
open circuit voltage), induced by the difference in electro-
chemical potential of electrons at both solid/gas interfaces,
develops across the oxide due to the redox equilibrium at the
interfaces. Expanding Wagner’s treatment
16
into the ternary
system of present concern, YSZ:N [or (Zr
1x
Y
x
)O
2x/23y/2
N
y
],
one can find that the open cell voltage of the cell
ðm0
O2;m0
N2Þ;PtjYSZ:NjPt;ðm00
N2;m00
O2Þ, under chemical potential
differences of oxygen and nitrogen is given as
E¼1
FZm00
O2
m0
O2
tO2
4dmO2þZm00
N2
m0
N2
tN3
6dmN2
0
@1
A;ð2Þ
where m
j
and t
k
denote the chemical potential of the thermo-
dynamic components j (j = O
2
,N
2
) and the transference
number of the ionic carriers k (k = O
2
,N
3
), and Fis the
Faraday constant. One can recognize in eqn (2) that the open
cell voltage, E, is a measure of the transference number of the
a
Institute of Physical Chemistry, Justus-Liebig-University,
Heinrich-Buff-Ring 58, D-35392 Giessen, Germany.
E-mail: [email protected]ie.uni-giessen.de;
Fax: +49-641-99-34509; Tel: +49-641-99-34500
b
TU Berlin, Institut fu
¨r Chemie, Straße des 17. Juni 135,
D-10623 Berlin, Germany
wCurrent address: Solar Cell Center, Korea Institute of Science and
zCurrent address: Institute for Solid State Research Electronic
Materials (IEM), Research Centre Juelich, D-52425 Juelich, Germany
PAPER www.rsc.org/pccp |Physical Chemistry Chemical Physics
Published on 16 November 2010. Downloaded by TU Berlin - Universitaetsbibl on 01/04/2016 08:06:19.
View Article Online
/ Journal Homepage
/ Table of Contents for this issue
1240 Phys. Chem. Chem. Phys., 2011, 13, 1239–1242 This journal is cthe Owner Societies 2011
mobile ionic species, and that the partial contributions of the
chemical potential differences of oxygen and nitrogen to Eare
independent of each other. In particular, when the chemical
potentials of nitrogen at both sides are equal, i.e.,m0
N2¼m00
N2,
one can determine the transference number of oxygen t
O
2
by
measuring Eas a function of oxygen activity at one side e.g.,
a00
O2½¼expðm00
O2=RTÞ while a0
O2at the other side is kept
constant, or
tO2¼4F
RT
@E
@lna00
O2a0
O2¼fixed
:ð3Þ
3. Experimental
Single crystals of YSZ:N were prepared by the nitridation
of YSZ (with 9.5 mol% Y
2
O
3
) single crystals grown by
the skull-melting technique. The nitridation was carried
out in a graphite heated resistance furnace (FCT-FSW 100/
150-2200-LA/PS) by thermal treatment in a nitrogen atmosphere
(1.06 10
5
Pa) at 1500 1C for 8 h. The nitrogen content,
determined by a Leco FE 300/400 analyzer, was 1.9 wt%.
Details of the nitridation and characterization of the samples
are presented elsewhere.
17
A single crystal of nitrogen-free
YSZ (CrysTec GmbH, Berlin) was also employed for the
purpose of comparison with the N-doped specimen, as well
as for confirming the validity of the present experiment.
For the measurement of open circuit voltage under an
oxygen activity gradient, a galvanic cell was constructed with
the configuration
Pt;a0
O2;a0
N2jðZr1xYxÞO2x=23y=2Nyja00
N2;a00
O2;Pt;ðIÞ
as illustrated in Fig. 1. A disk specimen measuring 20 mm
diameter 2 mm thickness was cut out of a single crystal
bowl, and its planar surfaces were polished down to 1 mm grit
diamond. A pair of gas electrodes was then formed on both
surfaces by the screen printing of platinum paste (Ferro GmbH,
no. 6402 1001). A piece of platinum gauze of ca. 0.25 cm
2
(Chempur GmbH, No 900338), which had been previously
welded to a platinum lead wire of 0.3 mm diameter (Chempur
GmbH, No 903695), was attached at an appropriate position
on each surface of the disk under a light spring pressure
applied through the alumina sheath carrying platinum wire.
The gas-tightness of each compartment of the cell was
achieved with the aid of a gold ring (20 mm outer diameter
1 mm thickness) mechanically pressed between the specimen
and each alumina tube. A gas mixing system was used to
separately control the activities of oxygen and nitrogen in each
compartment. The specimen was first equilibrated at a fixed
temperature in uniform activities of oxygen and nitrogen
(a0
O2¼a00
O2and a0
N2¼a00
N2) by using an N
2
/CO
2
/CO gas mixture
with a ratio of 10 : 70 : 20. The open circuit voltage, E, was then
measured against a00
O2in the range 20 ologa00
O2o14,
achieved by a stepwise variation of the mixing ratio 10 : x:y
(such that x+y= 90 and 50 rxr87) of the N
2
/CO
2
/CO
mixture while keeping a0
O2fixed at the initial value. The
nitrogen activities a0
N2and a00
N2were also fixed throughout
the experiment by fixing the mixing ratio of nitrogen gas to
0.1. The oxygen activities of the gas mixtures in both compart-
ments of the cell were monitored with additional independent
calcia-stabilized zirconia oxygen sensors (ex situ), one side of
which was exposed to air. The gas tightness of the cell was
confirmed, as shown in Fig. 2(a), by a stable oxygen activity
a0
O2on the left-hand side of the cell for a prolonged time upon
an abrupt exchange of gas composition on the right-hand side.
The open cell voltage measurement was carried out at different
temperatures in the range of 750 rT/1Cr850 while keeping
the sample and the oxygen sensors at the same temperature.
The temperature of the specimen, which was monitored with
two K-type thermocouples placed on both surfaces, was
controlled to within 11C.
4. Results and discussion
Fig. 2 shows the time variation of emf Eof galvanic cells with
(a) undoped YSZ and (b,c) YSZ:N, respectively, along with
Eof independent oxygen sensors upon stepwise changes
of oxygen activity a00
O2on the right-hand side of the cell
ða0
O2;a0
N2Þ;PtjspecimenjPt;ða00
N2;a00
O2Þwith a0
O2and a0
N2ð¼ a00
N2Þ
fixed. It is noted that each y-axis for the emf of the
oxygen sensor in Fig. 2 and Fig. 3 has the same signal window
(i.e. 200 mV in Fig. 2 and 250 mV in Fig. 3) as that for the emf
of the specimen. In Fig. 2(a), one may first recognize that the
change in emf, DE
YSZ
, of the nitrogen-free YSZ is almost
equal to that of the oxygen sensor for every step change
of a00
O2. The observed behavior of DE
YSZ
/DE
Sensor,R
E1isas
expected, since both the YSZ specimen and oxygen sensor
have an oxygen transference number of 1, thus supporting the
validity of the present measurement. On the other hand, the
emf E
YSZ:N
of the cell with N-doped YSZ behaves in a
different way, as shown in Fig. 2(b); DE
YSZ:N
/DE
Sensor,R
gets
smaller with decreasing E
Sensor,R
or a00
O2from the highest value
corresponding to the gas mixture N
2
/CO
2
/CO = 10/87/3,
which is more clearly seen in Fig. 2(c) for the result at a
different temperature of 850 1C.
The steady state values of E
YSZ:N
and E
Sensor,R
are plotted
as a function of loga00
O2in Fig. 3. Compared to the ideal slope
2.303 RT/4F of E
Sensor,R
vs. loga00
O2, the variation of E
YSZ:N
against oxygen activity gets less steep with decreasing
oxygen activity, which unambiguously demonstrates that the
Fig. 1 Schematic of the galvanic cell employed: 1, Al
2
O
3
tubing; 2,
gold ring gasket; 3, K-type thermocouple.
Published on 16 November 2010. Downloaded by TU Berlin - Universitaetsbibl on 01/04/2016 08:06:19.
View Article Online
This journal is cthe Owner Societies 2011 Phys. Chem. Chem. Phys., 2011, 13, 1239–1242 1241
transference number of oxygen in YSZ:N decreases with
decreasing a
O
2
due to eqn (3).
The transference number of oxygen in N-doped YSZ,
t
O
2
, was thus evaluated from the instantaneous slopes of the
E
YSZ:N
isotherms as a function of oxygen activity at different
temperatures, as shown in Fig. 4. The transference number,
t
O
2
, turned out to take values of less than 1 in the range of
log a00
O2o16 at 800 1C, suggesting that the transference
numbers of other types of charge carriers become significant
since P
k
tk¼1. The cation diffusivities of Zr and Y in YSZ:N
are much smaller than that of oxygen, and hence their contri-
bution can be safely excluded.
18
The electronic transference
number of YSZ is at most 10
5
in the range 19 ologa
O
2
o15
at 800 1C.
19
Even if it is generally expected that the electronic
partial conductivity, s
el
, is enhanced in N-doped YSZ due to
the additional N-states within the band gap,
20
s
el
in YSZ:N
has also been found to be negligibly small in comparison
with the total conductivity in the comparable experimental
conditions of present concern.
21
Consequently, we assign the
Fig. 2 Open circuit voltage E vs. t measured with galvanic cell
a0
O2;a0
N2jYSZ:N ðor YSZÞja00
O2;a00
N2along with E
Sensor
of zirconia-
based oxygen sensors: (a) E
YSZ
of undoped YSZ for a dummy test
at 800 1C, and E
YSZ:N
of N-doped YSZ at (b) 800 1C and (c) at 850 1C.
The a0
O2value on the reference side was kept invariant at each
temperature using a CO
2
/CO gas mixture with a fixed mixing
ratio of 70 : 20. a
N2
on both sides was also kept constant throughout
the measurement by flowing 10 sccm N
2
to each compartment.
For controlling the experimental variable a00
O2at the test electrode
(right-hand side) of the cell (I), CO
2
/CO mixtures with a mixing ratio
of 87 : 3 to 50 : 40 were employed.
Fig. 3 The open circuit voltage, E
YSZ:N
, of N-doped YSZ (closed
symbols) and the E
Sensor
of the oxygen sensor (open symbols) against
oxygen activity at different temperatures: , 850 1C; K, 800 1C; ,7501C.
Fig. 4 The transference number of nitrogen (t
N
3
, closed symbol) and
oxygen (t
O
2
, open symbol) in N-doped YSZ at different temperatures:
, 850 1C; K, 800 1C; , 750 1C. The solid curves are for the visual
guidance only.
Published on 16 November 2010. Downloaded by TU Berlin - Universitaetsbibl on 01/04/2016 08:06:19.
View Article Online
1242 Phys. Chem. Chem. Phys., 2011, 13, 1239–1242 This journal is cthe Owner Societies 2011
difference 1 t
O
2
to the transference number of nitrogen t
N
3
.
The results are also shown in Fig. 4. At very low oxygen
activities, YSZ indeed becomes electronically conducting.
22
At a given temperature, the nitrogen transference number of
YSZ:N increases from t
N
3
E0 with decreasing a
O
2
and
appears to be limited to a value of ca. 0.1 for the further
decrease in a
O
2
. As the oxygen activity decreases, the equilibrium
of the heterogeneous nitrogen incorporation reaction shifts
towards the right-hand side of eqn (1), thus resulting in the
increase of t
N
3
, as is observed in Fig. 4. Furthermore, it is also
recognized that t
N
3
takes a higher value with increasing
temperature at a fixed oxygen activity. The activation enthalpy
of oxygen diffusion in YSZ is known to be DH
D,O
= 0.7 to
1 eV, depending on the composition, while that of nitrogen
diffusion in YSZ:N is more than two times higher: DH
D,N
=2
to 2.5 eV.
6
As a consequence, the higher activation enthalpy of
nitrogen diffusion leads to an increasing t
N
3
with increasing
temperature. Also, the nitrogen incorporation reaction,
eqn (1), is endothermic, which is quite plausible considering
that the binding energy of N
2
(g) is higher than that of O
2
(g).
Consequently, YSZ accommodates more nitrogen in its lattice
at higher temperatures. According to nitridation studies, the
maximum nitrogen content at 2000 1C is approximately 15%.
3
The time evolution of the open circuit voltage of the
galvanic cell in Fig. 2 represents the progress of the surface
equilibration, not that of the bulk, upon an abrupt change of
oxygen activity in the gas phase. Once the surface of the oxide
is equilibrated, the emf does not change with time, regardless
of the degree of re-distribution of the non-stoichiometry in the
bulk. In terms of the kinetics of the emf change, there is little
noticeable difference between the undoped YSZ and the
N-doped YSZ. Thus, one may deduce that the heterogeneous
catalytic activity of the N-doped YSZ is no worse than that of
YSZ. In addition, changes in nitrogen transference number
upon a shift of oxygen activity, as shown in Fig. 4, suggests
that the concomitant shift of the thermodynamic equilibrium
of eqn (1) occurs, and hence nitrogen is electrochemically
active at the surface of YSZ:N. We noted that the YSZ:N
crystals—depending on the time and the number of experiments—
lost nitrogen, and thus changed their transference numbers
and resulting emf. The oxygen activity must always be kept
low in order to avoid the loss of nitrogen. However, we are
reminded that DE
YSZ:N
/DE
Sensor,R
gets smaller with decreasing
oxygen activity (Fig. 2b–c), implying the gradual enrichment
of nitrogen at the test electrode, and that the emf changes of
YSZ:N are reversible, at least during all the isothermal
measurements, as shown in Fig. 2b. Hence, we assign the
measured transference number to the presumably equilibrated
part of the specimen close to the electrode, despite ambiguity
over the stability of nitrogen in YSZ. We must point out that
future work needs to focus on the equilibrium concentration of
nitrogen in YSZ at lower temperatures than those yet studied
in order to obtain a deeper understanding of the electrode
properties and the transference of nitrogen.
5. Summary and conclusion
We have found that the nitrogen transference number in
YSZ:N varies from 0 to 0.1 at oxygen activities in the range
of 20 ologa
O
2
o14 at temperatures in the range of
1023 oT/K o1123, and the emf change of galvanic cells with
YSZ:N is as fast as that with YSZ. Whether the non-zero
nitrogen mobility together with the electrochemical activity of
nitrogen at the oxide/gas interface will find practical applica-
tions, e.g., in nitrogen sensors, nitrogen and/or ammonia fuel
cells, etc., remains questionable, as YSZ:N is not long-term
stable in an oxidizing atmosphere.
Acknowledgements
This study (DFG Ja648/ 8-3, Le781/ 10-4) was funded by the
German Science Foundation (DFG) within the priority
program 1136 Substitution Effects in Ionic Solids.
References
1 Y. B. Cheng and D. P. Thompson, J. Am. Ceram. Soc., 1993, 76,
683–688.
2 M. Lerch, J. Am. Ceram. Soc., 1996, 79, 2641–2644.
3 M. Lerch, J. Lerch and K. Lerch, J. Mater. Sci. Lett., 1996, 15,
2127–2129.
4 M. Lerch, J. Lerch, R. Hock and J. Wrba, J. Solid State Chem.,
1997, 128, 282–288.
5 J. Wendel, M. Lerch and W. Laqua, J. Solid State Chem., 1999,
142, 163–167.
6 M. Kilo, M. A. Taylor, C. Argirusis, G. Borchardt, M. Lerch,
O. Kaitasov and B. Lesag, Phys. Chem. Chem. Phys., 2004, 6,
3645–3649.
7 M. A. Taylor, M. Kilo, C. Argirusis, G. Borchardt, I. Valov,
C. Korte, J. Janek, T. Roedel and M. Lerch, Defect Diffus. Forum,
2005, 237–240, 479–484.
8 I. Kaiser-Bischoff, H. Boysen, C. Scherf and T. Hansen, Phys.
Chem. Chem. Phys., 2005, 7, 2061–2067.
9 J.-S. Lee, M. Lerch and J. Maier, J. Solid State Chem., 2006, 179,
270–277.
10 I. Valov, R. A. De Souza, C. Z. Wang, A. Boerger, C. Korte,
M. Martin, K.-D. Becker and J. Janek, J. Mater. Sci., 2007, 42,
1931–1941.
11 M. Lerch, J. Janek, K. D. Becker, S. Berendts, H. Boysen,
T. Bredow, R. Dronskowski, S. G. Ebbinghaus, M. Kilo,
M. W. Lumey, M. Martin, C. Reimann, E. Schweda, I. Valov
and H. D. Wiemho
¨fer, Prog. Solid State Chem., 2009, 37,
81–131.
12 T.-J. Chung, J.-S. Lee, D.-Y. Kim and H. Song, J. Am. Ceram.
Soc., 1999, 82, 3193–3199.
13 G. Deghenghi, T.-J. Chung and V. Sergo, J. Am. Ceram. Soc.,
2003, 86, 169–173.
14 I. Valov, C. Korte, R. A. De Souza, M. Martin and J. Janek,
Electrochem. Solid-State Lett., 2006, 9, F23–F26.
15 I. Valov and J. Janek, Solid State Ionics, 2006, 177,
1619–1624.
16 C. Wagner, Z. Phys. Chem., 1933, B21, 25–41.
17 T.-C. Roedel, D. Wang, D. S. Su, M. Lerch, A. Prokofiev,
K. D. Luther and W. Assmus, Cryst. Res. Technol., 2006, 41,
950–954.
18 M. Kilo, M. A. Taylor, C. Argirusis, G. Borchardt, B. Lesage,
S. Weber, S. Scherrer, M. Schroeder and M. Martin, J. Appl. Phys.,
2003, 94, 7547–7552.
19 J.-H. Park and R. N. Blumenthal, J. Electrochem. Soc., 1989, 136,
2867–2876.
20 T. Bredow, Phys. Rev. B: Condens. Matter Mater. Phys., 2007, 75,
144102.
21 I. Valov, V. Ru
¨hrup, R. Klein, T.-C. Ro
¨del, A. Stork, S. Berendts,
M. Dogan, H.-D. Wiemho
¨fer, M. Lerch and J. Janek, Solid State
Ionics, 2009, 180, 1463–1470.
22 J. Janek and C. Korte, Solid State Ionics, 1999, 116,
181–195.
Published on 16 November 2010. Downloaded by TU Berlin - Universitaetsbibl on 01/04/2016 08:06:19.
View Article Online
