Ionic and electronic conductivity of nitrogen-doped YSZ single crystals

Ionic and electronic conductivity of nitrogen-doped YSZ single crystals

Solid State Ionics 180 (2009) 1463–1470 Contents lists available at ScienceDirect Solid State Ionics j o u r n a l h o m e p a g e : w w w. e l s ev...

787KB Sizes 2 Downloads 52 Views

Solid State Ionics 180 (2009) 1463–1470

Contents lists available at ScienceDirect

Solid State Ionics j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / s s i

Ionic and electronic conductivity of nitrogen-doped YSZ single crystals I. Valov a,⁎, V. Rührup d, R. Klein b, T.-C. Rödel c, A. Stork c, S. Berendts c, M. Dogan d, H.-D. Wiemhöfer d, M. Lerch c, J. Janek b,⁎ a

Institute for Solid State Research, Electronic Materials, Science Centre Juelich, 52425 Juelich, Germany Institute of Physical Chemistry, Justus-Liebig University, Heinrich-Buff-Ring 58, 35392 Giessen, Germany Institute of Chemistry, Technical University Berlin, Straße des 17. Juni 135, 10623 Berlin, Germany d Institute of Inorganic and Analytical Chemistry, Westfälische Wilhelms-Universität, Corrensstr. 30, 48149 Münster, Germany b c

a r t i c l e

i n f o

Article history: Received 4 May 2009 Received in revised form 2 September 2009 Accepted 11 September 2009 Keywords: Nitrogen electrochemistry Ionic conductivity Electronic conductivity Nitrogen doping YSZ Hebb–Wagner measurements

a b s t r a c t The ionic and electronic charge transport was studied for single crystals of 9.5 mol% yttria-stabilized zirconia with additional nitrogen doping (YSZ:N) of up to 7.5 at.% (referred to the anion sublattice and formula unit Zr0.83Y0.17O1.91) as a function of temperature and nitrogen content. The total conductivity being almost equivalent to the oxygen ion conductivity has been measured by AC impedance spectroscopy under vacuum conditions in order to prevent re-oxidation and loss of nitrogen. The electronic conductivity has been determined by Hebb–Wagner polarization using ion-blocking Pt microelectrodes in N2 atmosphere. The ionic conductivity of YSZ:N decreases in the presence of nitrogen at intermediate temperatures up to 1000 °C. The mean activation energy of ionic conduction strongly increases with increasing nitrogen content, from 1.0 eV for nitrogen-free YSZ up to 1.9 eV for YSZ containing 7.3 at.% N. Compared to nitrogen-free YSZ, the electronic conductivity first decreases at nitrogen contents of 2.17 and 5.80 at.%, but then increases again for a sample with 7.53 at.%. At temperatures of 850 °C and above, the presence of the N3− dopant fixes the electrode potential and thus the oxygen partial pressure at the Pt electrode to very low values. This corresponds to a pinning of the Fermi level at a relatively high energy in the upper half of the band gap. At 7.53 at.% N and 950 °C, the oxygen partial pressure in YSZ:N corresponds to pO2 = 3 × 10− 18 bar. At temperatures above 850 °C, even in the presence of a very small oxygen concentration in the surrounding gas phase, the nitrogen ion dopant becomes highly mobile and thus diffuses to the surface where it is oxidized to gaseous N2. The results are discussed in terms of the ionic and electronic defect structures and the defect mobilities in YSZ:N. © 2009 Elsevier B.V. All rights reserved.

1. Introduction It is well known from a series of studies that nitrogen can be incorporated into zirconia-based solid electrolytes [1–6]. Different approaches were applied to incorporate nitrogen into YSZ — high temperature treatment [7,8], electrochemical incorporation [9] and pulsed laser deposition (PLD) [10] with a maximum concentration of 2.4 wt.% nitrogen in YSZ as reported by Lerch [6,7]. The incorporation into pure ZrO2 leads to the cubic γ-phase or to ordered β-type-phases with a lower symmetry [4,11]. The incorporation of nitrogen into partially or fully yttria-stabilized zirconia (YSZ) is known to be an

⁎ Corresponding authors. Valov is to be contacted at Institute for Solid State Research, Electronic Materials, Science Centre Juelich, 52425 Juelich, Germany. Janek, Institute of Physical Chemistry, Justus-Liebig University, Heinrich-Buff-Ring 58, 35392 Giessen, Germany. J. Janek is to be contacted at Tel.: +49 641 9934500; fax: +49 641 9934509. I. Valov, Tel.: +49 2461 612994; fax: +49 2461 612550. E-mail addresses: [email protected] (J. Janek), [email protected] (I. Valov). 0167-2738/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.ssi.2009.09.003

alternative route for the creation of anion vacancies as compared to the conventional doping with lower valent cations according to ′

••

x

ð1Þ

Y2 O3 = 2YZr + VO + OO

The corresponding effect of a substitution of oxygen by nitrogen is characterized by the formation of further anion vacancies according to x



••

N2 ðgasÞ + 3OO ðYSZÞ = 2NO ðYSZÞ + VO + 3 = 2O2 ðgasÞ

ð2Þ

Therefore, in the same way as the doping with lower valent cations, the substitution of oxygen ions by nitrogen ions stabilizes the cubic fluorite-type structure due to the enhanced oxygen vacancy concentration. Nitrogen-doped YSZ in both cubic and tetragonal structures was found to be an excellent ionic conductor showing even nitrogen ion conductivity at high temperatures [8,12–15]. However, Eq. (2) also implies that an increasing oxygen partial pressure supports nitrogen loss by oxidation, if nitrogen ions are mobile in the solid. Accordingly, applications of the special properties of nitrogen-doped YSZ and its

1464

I. Valov et al. / Solid State Ionics 180 (2009) 1463–1470

nitrogen ion conductivity are limited at high temperatures to low oxygen partial pressures. Lerch et al. demonstrated that anion (nitrogen) substitution in ZrO2 or in tetragonal yttria doped zirconia (Y-TZP) stabilizes the cubic CaF2-type structure and improves the transport properties of these materials [3,4,6,12,16]. Nitrogen doping is also reported to result in a partially filling of the conduction band and is therefore applied to tune the electronic properties in oxide semiconductors, e.g. in TiO2 [17]. The general aim of our studies on nitrogen doping of oxides is twofold — firstly a better understanding of the change in the defect structure and the thermodynamics of the oxide ceramics after addition of N3− in the anion sublattice, and secondly, to achieve sufficient nitrogen ion conductivity for the development of a nitrogen sensor or nitrogen pumps. In respect to the latter, several experimental studies on the electrochemical surface kinetics of dinitrogen reduction and on the transport properties were published, and as well theoretical calculations on the defect formation energies have been performed [1,13,14,18–21]. We demonstrated that the surface kinetics and not the diffusion into the bulk limit the overall process at a temperature of about 700 °C [20]. Nitrogen-doped YSZ single crystals show a small partial nitrogen ion conductivity at temperatures above 400 °C with a diffusion coefficient of nitrogen ions determined by tracer diffusion experiments of D(NO′) ≈2 ×10− 11 cm2 s− 1 at 700 °C [19] with an activation energy close to 2 eV. These numbers coincide well with the results published by other authors on polycrystalline materials [1,13,14,18]. The electronic conductivity of polycrystalline yttria doped zirconia was measured by many authors (cf. values reported for temperatures at and above 800 °C in [22,23]). It is rather low with a minimum at around p(O2)10− 10 bar and 950 °C. Data at lower temperatures are missing. A reason is that very small electronic conductivities far below 10− 6 Ω− 1 cm− 1 occur at temperatures below 900 °C. Nevertheless, it is clear from the literature that the intrinsic n- and p-type contributions to the electronic conductivity of YSZ follow an oxygen partial pressure dependence given by

σe =

∘ σn

pO 2

!−1 = 4

pO∘ 2

+

∘ σp

pO 2

!

pO∘ 2

+ 1=4

many) at temperatures from 1500 °C up to 1900 °C and reaction times between 2 and 12 h in nitrogen atmosphere (1.01 × 105 Pa). Nitrogen contents were determined using a Leco EF-TC 300 (Leco, Saint Joseph, USA) hot gas extraction analyzer. The thin rock salt-type (Y,Zr)(O,N,C) surface layer (thickness up to ~100 μm, depending on nitridation conditions) was removed before chemical analysis and impedance spectroscopy measurements were carried out. As function of the reaction parameters the bulk nitrogen content was found to be up to 2.6 wt.%. Respecting the substitution mechanism presented in Eq. (2) a stoichiometry of Y0.17Zr0.83O1.63N0.19 can be calculated for the crystal containing 2.6 wt.% nitrogen. In the following we refer the concentration of N3− and V•• O (in atomic percent) to the total sites in the anion sublattice in the cubic YSZ:N with stoichiometric formula N2ð2 + xÞ Vx + nð2 + xÞ , where x and n are Zr 1−x Y 2x O2 + x ð1−nÞ n 3 1+x 1+x 1+x

2. Experimental

1+x

2.2. Ionic conductivity measurements Two point impedance measurements were performed with a EG & G potentiostat (Model 283) equipped with a frequency response detector Model 1025 operating in the frequency range between 5 MHz and 1 mHz with a maximum AC amplitude of 100 mV. The conductivity of the samples was calculated according to

ð3Þ

where σn∘ and σp∘ stand for the n-type and p-type conductivities at the reference oxygen partial pressure pO° 2. The basic assumption for the validity of Eq. (3) is that the oxygen vacancy concentration is fixed by a high dopant concentration (usually two- or three-valent cations). In the case of nitrogen-doped YSZ, an effect on the electronic conductivity is difficult to predict. First, there is usually a decrease of the electron concentration when the acceptor dopant concentration is increased. On the other hand, the nitrogen doping requires a low oxygen activity which increases the electronic conductivity. Furthermore, nitrogen ions, if mobile, may be oxidized at the surface and then set free additional electrons, effectively leading to the reduction or decomposition of the solid phase. At high nitrogen ion concentrations, formation of an impurity band can lead to additional conducting electron states. In the following, we present data for the ionic as well as for the electronic conductivity of (111) and (100) oriented YSZ:N single crystals with varying nitrogen contents and discuss the results in terms of defect models.

3ð1 + xÞ

the fractions of the cation and anion dopant, respectively. In this study x is fixed to 0.095 and n is varied in the range 0–0.12. Details of nitridation are published in a recent contribution [7]. The typical dimensions of the crystals used for impedance measurements were rod-like of dimensions 10× 1 × 1 mm3. The crystals were electrically contacted at both sides with platinum paste (Ferro Corporate Headquarters, Cleveland, Ohio, USA), dried in air and prefixed at 200 °C for 2 h in vacuum. For the micro-contact measurements we used samples of typical dimensions 5 × 5 × 1 mm3, contacted to a Cu, Cu2O reference at the back side and a platinum micro-electrode at the front side. The oxygen reference was electrically contacted with platinum paste.

σtotal =

l RA

ð4Þ

where σtotal is the total conductivity, l is the distance between the two electrodes, A is the sample cross section (equal to the electrode area) and R is the resistance determined for the bulk semi-circle as evaluated from AC impedance data. The recorded impedance spectra were deconvoluted with the program EQUIVALENT CIRCUIT written by B. Boukamp. The temperature was measured by a thermocouple of type K at the sample surface with an experimental error of ±1 K. 2.3. Electronic conductivity measurements The electronic conductivity was measured according to the Hebb– Wagner polarization technique using a blocking platinum microcontact in the following electrochemical cell in the temperature range between 700 °C and 950 °C in a purified nitrogen atmosphere. To avoid moisture the gas was purged through a self assembled series of purification columns containing KOH, P4O10 and silica gel. The rest oxygen activity in the gas was measured by a lambda cell and was in the order of 10− 6. Cu; Cu2 Oj YSZ : N j PtðmicrocontactÞ

ð5Þ

2.1. Preparation Single crystals of YSZ (yttria-stabilized zirconia, 9.5 mol% Y2O3) were prepared by the skull melting method [24,25]. Thin plates in the dimensions 10 × 10 × 1 mm3 with the crystallographic orientations (100) and (111) were nitrided in a graphite-heated resistance furnace (FCT — FSW 100/150-2200-LA/PS, FCT-Anlagenbau, Sonneberg, Ger-

The use of a microcontact minimizes the relaxation time to reach a steady-state polarization after each voltage step. The carefully cleaned platinum microelectrodes (polished, flattened tip) were placed with slight pressure on top of a polished planar crystal surface of the YSZ:N or YSZ single crystals. Measurements on the (100) oriented single crystals yielded the best signal/noise ratio and therefore were taken for this

I. Valov et al. / Solid State Ionics 180 (2009) 1463–1470

work. The radius of the circular microcontact area was typically in the range of 30–70 μm. The opposite side of the single crystals was contacted by a planar reference electrode made of an annealed Cu, Cu2O mixture which served to fix the oxygen activity and usually covered the entire back surface. The oxygen partial pressure pO⋆ 2 at the reference as a function of temperature was taken from published data [26] giving ⋆

log10 ½pO 2 = 1:013 bar = 7:5159−

17486 : T=K

ð6Þ

The steady-state cell voltage U, i.e. the electrode potential of the microelectrode vs. the potential at the large area reference contact is given by the Nernst equation under the assumption of small currents and negligible contact resistance U=

! pO RT ln ⋆ 2 : pO 2 4F

ð7Þ

pO2 is fixed by the applied voltage and denotes the local oxygen partial pressure at the interface between YSZ:N (or YSZ) and the microelectrode. In the case of a non-negligible contact resistance Rcont, the voltage used in Eq. (7) has to be taken from the measured cell voltage Ucell according to U = Ucell −Rcont I:

ð8Þ

The local electronic conductivity σe (pO2) is then calculated from the slope of the steady-state current–voltage curve according to σe ðpO 2 Þ =

  1 ∂I πd ∂U UðpO

2

Þ

ð9Þ

where d corresponds to the diameter of the contact area at the microelectrode. An extensive theoretical treatment of the basic assumptions for the Hebb–Wagner evaluation with microcontacts has been given elsewhere [27–29]. 3. Results and discussion 3.1. Total AC conductivity of YSZ:N single crystals 3.1.1. Measurements during the heating cycles Impedance spectra at different temperatures of an YSZ:N (100) single crystal are shown in a Nyquist plot in Fig. 1. The individual

Fig. 1. Nyquist plot of measured AC impedance of 9.5YSZ:N (100) single crystal at different temperatures in the range from 600 °C to 800 °C.

1465

spectra consist of a complete semi-circle (high frequency range), representing the bulk transport, and a part of a second semi-circle, representing the electrode impedance (as the contacting electrodes were symmetric, they are represented by one semi-circle) appears at low frequencies. With increasing temperature the experimental dispersions shift towards higher frequencies and the resistance of the sample decreases. The first sub-circuit consists of a resistance Rbulk and capacitance Cbulk in parallel representing the bulk transport properties. The second sub-circuit is composed of a resistance Relectrode and a constant phase element (CPE), connected in parallel. The representative fit given in Fig. 2 showed that the bulk resistance was lower by about one order of magnitude as compared to the electrode resistance. We also found that the power n of the constant phase element (the impedance of the constant phase element is given by Z = (A(iω)n)− 1, where A and n are fit parameters) is close to 1, indicating that the electrode capacity is nearly ideal. The total conductivities of YSZ:N single crystals ((111) and (100) oriented samples) with different nitrogen contents measured as a function of the temperature, resembled those reported by Wendel et al. for polycrystalline samples [12] and were in general lower compared to the conductivity of nitrogen-free YSZ crystals. Arrhenius plots for the first heating cycle are shown in Figs. 3 and 4. The total (ionic) conductivity decreases with increasing nitrogen content, and this effect is pronounced at temperatures below 1000 °C in spite of the fact that nitrogen doping is expected to increase the concentration of anion vacancies (according to Eq. (2)). The situation changes to the opposite above 1000 °C where the nitrogen-doped samples show higher ionic conductivities than the nitrogen-free YSZ. At this point we should mention that, in spite of the dark color, the single crystals of 9.5YSZ:N showed a high ohmic resistance. Studies on the reduction (blackening) of YSZ single crystals have confirmed that an increase of the electronic conductivity can be expected at extremely high voltages, respectively low oxygen activities [30]. However, the DC polarization at blocking electrodes always yielded an electronic conductivity being orders of magnitude lower compared to the ionic conductivity (Figs. 12 and 13). Thus, the measured total conductivity is indeed predominantly ionic. However, a distinction between oxygen and nitrogen conductivities based on the present experiments is not possible. In the Arrhenius plots, two different regions with different activation energies can be distinguished at lower temperature (250 °C–650 °C) and at higher temperature (650 °C up to 950 °C). The effects of the different nitrogen contents on the activation energies and of the expected increase of the concentration of oxygen vacancies (calculated) on the conductivity are presented in Figs. 5 and 6. The activation

Fig. 2. Equivalent circuit of the AC electric response of (100) 9.5YSZ:N single crystal based on the fit procedure to the experimental impedance spectra.

1466

I. Valov et al. / Solid State Ionics 180 (2009) 1463–1470

Fig. 5. Dependence of the total bulk conductivity of (100) YSZ:N (empty symbols) and (111) YSZ:N (filled symbols) on the calculated total vacancy concentration. Fig. 3. Arrhenius plots of total conductivity of the YSZ:N (111) single crystals with a nitrogen content up to 7.3 at.%. The measurements were performed in the direction of increasing temperature. The nitrogen-free sample was (100) oriented.

energies are higher for samples with higher nitrogen content. The low temperature range is characterized by higher activation energies (up to 30%) compared to the high temperature region. Neither a quantitative nor a qualitative difference in the transport properties for the single crystals with different orientations was observed. We attribute the observed lower conductivity and higher activation energies to both local stress and coulomb interactions between the N3− ions and V•• O. The nitrogen ion is characterized by a slightly larger ionic radius compared to oxygen (Δr = 8 pm for coordination number 4 [31,32]) and bears an additional negative charge, in the defect nomenclature denoted by NO′. These factors will be responsible at low temperatures for the decrease of the ionic conductivity and the increasing activation energy which probably then includes an association enthalpy. In addition N3− and V•• O can build clusters effectively trapping the vacancies, thus decreasing the ionic conductivity (N3− and O2−). This effect was reported for nitrogen-free YSZ demonstrating that at temperatures around 700 °C the oxygen vacancies (trapped at lower ⁗, or Y′Zr) temperatures in clusters by negatively charged defects like VZr become more mobile, i.e. the thermal energy is enough to activate their free motion. In addition at temperatures above 780 °C, yttria as well as

Fig. 4. Arrhenius plot of the total conductivity of YSZ:N (100) single crystals with a nitrogen content up to 7.3 at.%. The measurements were performed in the direction of increasing temperature.

some other impurities segregates to the surface and leads to further influence on the conductivity [33,34]. We assume for our samples that at temperatures higher than approx. 900 °C the nitrogen (and also yttrium ion) association clusters dissociate in full releasing the oxygen vacancies and increasing the conductivity, respectively. One can conclude that the concentration of defect associates which lead to lower vacancy mobilities at the high temperatures becomes small. Then, the main influence of the increased vacancy concentration due to nitrogen doping becomes dominant. At the same time, the freed nitrogen ions appear to become mobile as the observed loss of nitrogen showed. These facts can be correlated well with observations of other authors who found that from 400 °C to 800 °C the nitrogen diffusion coefficients were at least three orders of magnitude lower than the diffusion coefficient of oxygen in the same range [18,19]. But, at temperatures higher than 900 °C, this difference in diffusion of the two anion species became small and an extrapolation of the experimental data even suggested that at temperatures above 1000 °C both diffusion coefficients attain the same order of magnitude. Thus, the increase of the ionic conductivity around 1000 °C is accompanied by an enhanced nitrogen ion mobility via vacancies which accordingly also contribute to the total conductivity.

Fig. 6. Activation energies for the conductivity of YSZ:N (100) (empty symbols) and YSZ:N (111) (filled symbols) single crystals as a function of nitrogen content. The two lines represent mean data for the two different temperature regions. The vertical dot lines mark the position of the n value.

I. Valov et al. / Solid State Ionics 180 (2009) 1463–1470

To conclude, in the presence of nitrogen ions, the association of vacancies and both negatively charged acceptor dopants seems to depress the mobility of vacancies at lower temperatures. 3.1.2. Measurements during the cooling cycles The conductivity measurements of the samples were repeated during the cooling cycle in order to study the stability of the YSZ:N crystals with respect to a possible re-oxidation (a loss of nitrogen). As it can be seen in Fig. 7, with an exception of the YSZ:N crystal doped by 6.2 at.% nitrogen, the conductivities of all samples increase and are of the same value irrespective of the initial nitrogen content. This clearly indicates a re-oxidation of the samples and a corresponding nearly complete loss of nitrogen at the former high temperatures. The nitrogen loss was experimentally proven by hot gas extraction analysis on the chemical composition of the YSZ:N crystals after the conductivity measurements (after a complete heating/ cooling cycle). The analysis shows that all samples (being of white color) were completely re-oxidized. Obviously, in spite of the vacuum conditions (p(O2) < 10− 4 mbar), the residual traces of oxygen seem to be enough to replace the nitrogen in the high temperature range by re-oxidation (average duration of the measurement was one week). These observations make the high nitrogen ion mobility at elevated temperatures evident. The only exception was the sample initially doped with 6.2 at.% nitrogen which had slightly reduced (and not lost) its nitrogen content down to 5.8 at.%. Visually this crystal was of dark color in the middle with a transparent to white overlayer. However, we have no explanation why this particular YSZ:N sample doped with 6.17 at.% nitrogen behaved differently and did not re-oxidize completely. 3.2. Electronic conductivity of YSZ:N single crystals 3.2.1. Leakage currents due to non-capsulated microcontacts Usually, in earlier experiments on oxygen ion conducting solids [29], we applied a non-conducting glassy paste to ensure a gas tight encapsulation around the microelectrodes (cf. [29]). The glassy paste, however, has to be annealed before use at 1000 °C for 2 h in oxygen. This causes a conflict with the presence of the high nitrogen dopant concentration, as the oxidative high temperature heat treatment would remove the nitrogen dopant almost completely (due to reaction 2). Thus, in order to prevent fast oxidation and loss of the nitrogen, we decided to carry out the measurements in purified N2 atmosphere without encapsulating the microelectrodes. In the following, we briefly discuss the influence of this approach.

Fig. 7. Arrhenius-plot of the total conductivity of YSZ:N (100) single crystals with a nitrogen content up to 7.3 at.%. The data were obtained during the cooling cycle.

1467

For the highest positive cell voltages, i.e. U > 300 mV, the increasing oxygen partial pressure at the microcontact will result in the formation of gaseous oxygen and, therefore, an increasing ionic current. Accordingly, for higher positive cell voltage, the microcontact will be no longer blocking. Evaluation of the steady-state current–voltage curves in terms of the electronic conductivity according to Eq. (9) will then yield values that are too high. More advantageous conditions exist for negative cell voltages where the polarization voltage at the microcontact generates highly reducing conditions. Certainly, the small oxygen concentration remaining in the purified N2 atmosphere will lead to an additional reduction current at the microcontact corresponding to a small ionic current in the sample [35,36]. Results from earlier experiments on Pt microcontacts on YSZ give a good estimate of the magnitude [36]: negatively polarized microcontacts with a diameter of 140 μm showed diffusion limited oxygen currents in O2/N2 mixtures at 950 °C of 40 μA at 10− 2 bar oxygen partial pressure and 5 μA at 10− 3 bar. Extrapolating to an oxygen partial pressure of 10− 5 bar yields estimated reduction currents of around 50 nA. Fig. 8 shows the steady-state current–voltage curves obtained for a (100) oriented single crystal of YSZ without any nitrogen doping. It is evident that at 950 °C, the electronic current is considerably higher than those 50 nA. Furthermore, as the oxygen reduction below −200 mV occurs under limiting conditions, it is constant and thus does not contribute to the slope of the steady-state I–U curve and to the electronic conductivity calculated with Eq. (9). The I–U curves in Fig. 8 show a zero crossing at positive voltages between 50 mV and 150 mV. For an ideal Hebb–Wagner experiment, this should occur at 0 mV. The shifted zero is explained by the electrode equilibrium of the very small oxygen impurity in the applied nitrogen atmosphere. The oxygen partial pressures calculated from the zero-current voltages are on the average one to two orders of magnitude higher than that of the Cu, Cu2O reference (reference oxygen partial pressures are, for instance, 1.7∙10− 9 bar at 800 °C, and 4.0∙10− 8 bar at 900 °C). Fig. 9 shows the conductivities derived from the curves in Fig. 8 with the help of Eq. (9). It is evident that the slopes at oxygen partial pressures above 10− 5 bar are higher than the expected +1/4. This is due to the beginning anodic formation of gaseous oxygen in that range which occurs at the microcontact for positive electrode potentials (vs. the reference). As no glass sealing was applied, the ion-blocking property of the microcontact is best for negative electrode potentials. However, the flat region around the conductivity minimum in Fig. 9 has nothing to do with a non-blocking property of the microcontact. One also cannot explain it by a contact resistance as

Fig. 8. Steady-state current–voltage curves with Pt microcontact in N2 atmosphere on a (100) oriented single crystal of YSZ (9.5 mol% Y2O3) without N-doping (contact diameter of the microelectrode: 35 μm); reference electrode: Cu, Cu2O mixture).

1468

I. Valov et al. / Solid State Ionics 180 (2009) 1463–1470

Fig. 9. Electronic conductivity as a function of oxygen partial pressure, evaluated from the steady-state I–U curves of Fig. 8.

that would give the largest effect for the large currents to the right and left of the conductivity minimum (and would imply a considerable deviation from − 1/4 slope at low oxygen partial pressures which is not detected). We therefore think that this pinned electronic conductivity around the conductivity minimum is due to a small leakage current from impurities such as traces of transition metals. It is well known from publications of other authors (cf. [37,38]) that these cause deep electronic levels in the band gap (sometimes acting as traps for electrons) and become best detectable near the electronic conductivity minimum and for low temperatures. Accordingly, this electronic trap contribution gets much weaker for increasing temperatures being almost negligible at 950 °C.

3.2.2. Measurements on YSZ:N single crystals Fig. 10 shows steady-state current–voltage curves obtained for an (100) oriented crystal with 7.53 at.% nitrogen-dopant concentration at temperatures between 700 °C and 950 °C. The general appearance of the steady-state current–voltage curves (see also Fig. 8) follows the sigmoidal shape in agreement with earlier Hebb–Wagner experiments on polycrystalline YSZ [22,23]. The current plateau in the middle voltage range marks the minimum of the electronic conductivity where a change from n-type to p-type electronic conductivity

Fig. 10. Steady-state current–voltage curves with Pt microcontact in N2 atmosphere on a (100) oriented single crystal of YSZ (9.5 mol% Y2O3) with 7.53 at.% N-doping (contact diameter of the microelectrode: 36 μm; reference electrode: Cu, Cu2O mixture).

occurs for increasing potential of the microelectrode vs. the Cu, Cu2O reference. For temperatures of 850 °C and below, the I–U curves virtually have the same appearance as that found for the undoped YSZ in Fig. 8. Again, one finds the zero crossing of the curves at positive voltages between +100 mV and + 200 mV. However, a remarkable change was observed for the I–U curves on all N-doped samples, if the temperature increased to 900 °C and 950 °C. Then, all nitrogen-doped samples (2.17 at.%, 5.8 at.% and 7.53 at.% N) showed a distinct shift of the zero current point towards highly negative values (e.g. in Fig. 10, it is −450 mV at 900 °C, and − 640 mV at 950 °C). It is evident that the jump occurs at temperatures higher than 850 °C. It should be noticed that the shift of the zero current crossing to negative values correlates well with the temperatures where an increasing total conductivity and an accelerated loss of nitrogen was observed. At 950 °C, using Eq. (7) the microelectrode potential of −640 mV vs. the solid Cu, Cu2O reference corresponds to an equilibrium oxygen partial pressure of nearly 10− 18 bar. This means that the potential of the Pt microelectrode is fixed by an intrinsic redox reaction at the interface which we believe corresponds to the redox couple N3− (on O2− site)/N2(g). Between −640 mV and 0 mV, one observes a net oxidation current due to the redox activity of the nitrogen dopants which are quite mobile at temperatures above 900 °C. It is evident that the electrode reaction responsible for the highly negative potential is given by 1 ′ •• − NO ðYSZÞ⇌ N2 ðgasÞ + VO + 3e 2

ð10Þ

The observed oxidation current under negative cell voltages between −640 mV and 0 mV is due to a cell reaction driven by the surface loss of nitrogen and a corresponding supply of oxygen by the reference contact. One can assume that the corresponding current is rate limited by the out-diffusion of nitrogen ions from the bulk. The highly reducing conditions imply a superposition of the ionic nitrogen current and the exponentially voltage dependent electron current. Under such conditions, the limiting current contributes negligibly to the slope of the steady-state I–U curve and, therefore, the evaluation of the slope in terms of the electronic conductivity makes sense. The electronic conductivities evaluated for the 7.53 at.% sample from the curve slopes in Fig. 10 are plotted in Fig. 11 vs. the logarithm of the oxygen partial pressure. In the very low oxygen partial pressure range to the left of the conductivity minimum, the curves again nicely follow the −1/4 slope as expected for intrinsic conduction electrons in a predominating ionic conductor. In this range, the exponential

Fig. 11. Electronic conductivity as a function of oxygen partial pressure evaluated from the steady-state I–U curves for the 7.53 at.% N-doped YSZ of Fig. 10.

I. Valov et al. / Solid State Ionics 180 (2009) 1463–1470

increase of the electron concentration by far overtakes the current due to nitrogen oxidation. From the temperature dependance of the electronic conductivities at constant oxygen partial pressures in the n-type and p-type branches, the corresponding activation enthalpies ΔHA were calculated. At low oxygen partial pressures in the n-type range, the average value was ΔHA,n = 2.85 ± 0.15 eV, and in the p-type range at high oxygen partial pressures, i.e. to the right of the conductivity minimum, an average of ΔHA,p = 1.0 ± 0.15 eV was obtained. These values did not change for the other nitrogen concentrations. One can conclude that the electronic conductivity still is dominated by the intrinsic electron and hole conduction of YSZ. The calculated transference number of the electrons at pO2 = 10− 6 bar for entire temperature and nitrogen concentration range did not exceed 0.004. Figs. 12 and 13 give an overview over the dependence of the electronic conductivities on the nitrogen concentration for 850 °C and 950 °C. There is no change in the general appearance at these two temperatures, despite the high nitrogen mobility at the higher temperature. This comparison suggests that the nitrogen ion mobility does not directly contribute to the electronic transport. For the two lower nitrogen concentrations (2.17 and 5.8 at.%), the electronic conductivities are smaller as compared to the N-free YSZ, if one takes values at the same oxygen partial pressure. We suppose, that the vertical shift of the electronic conductivity curve towards lower values is a result of introduced scattering centers (nitrogen ions) lowering effectively the mobility of the electronic charge carriers. We measure by up to 0.7 eV higher activation enthalpies for the electronic transport in YSZ:N in comparison to nitrogen-free YSZ indicating that the energetic barrier for electron transport has been changed by the nitrogen doping. However, this hypothesis needs additional theoretical calculations to be supported (a subject of currently running activities). In contrary to this, at the highest N concentration, one observes a considerable jump of the entire electronic conductivity curve towards higher values reaching the level of the electronic conductivity of N free zirconia. We assume that this result reflects the onset of a change in the electronic state densities near the band edges due to the increasing substitution of oxide by nitrogen ions. It is caused by the increasing dispersion (band tailing) due to the steep increase of ionic defect concentrations (both vacancies and nitrogen ions). It should be noted that binary nitrides such as CeN and ZrN are very good metallic type electronic conductors. Therefore, there is no direct hint for the formation of an impurity band by nitrogen as that must lead to stronger conductivity enhancement. This is supported by

Fig. 12. Comparison of electronic conductivities measured for different N-concentrations of (100) oriented single crystals of 9.5YSZ:N at 850 °C with N contents of 0, 2.17, 5.8 and 7.53 at.%.

1469

Fig. 13. Comparison of electronic conductivities measured for different N-concentrations of (100) oriented single crystals of 9.5YSZ:N at 950 °C with N contents of 0, 2.17, 5.8 and 7.53 at.%.

the fact that the activation energies for hole and electron conductivity do not change as a function of the nitrogen concentration. Finally, taking quantitative data from an earlier publication with respect to the band scheme of YSZ and its relation to the electrode potential scale [39], the electrochemical observation of a nitrogen ion related electrode potential at 950 °C can be used to suggest the approximate position of the highest occupied electronic levels of nitrogen ions in the band gap of YSZ. Fig. 14 shows the result of this interpretation. As the highest occupied level of the nitrogen ions lies in the upper half of the band gap, it fixes the Fermi level to a high value lying 2.8 eV above the valence band edge. This is comparable to the position of the Fermi level in the presence of a H2/H2O mixture (p(O2) ~ 10− 20 bar). Accordingly, the electrode potential is fixed to a highly negative value as observed. Lowering the Fermi level would always imply a removal of electrons from the highest occupied nitrogen orbitals and accordingly an oxidation of nitrogen ions. 4. Conclusions Both, the ionic and the electronic conductivities of nitrogen-doped 9.5YSZ single crystals with (111) and (100) orientation showed

Fig. 14. Band scheme for YSZ and relative position of the additional occupied redox levels due to dissolved nitrogen ions and oxidized species Nx− with x < 3. εF denotes the Fermi energy. The additional scales illustrate the one-to-one relation between Fermi energy, oxygen partial pressure and oxygen electrode potential [39].

1470

I. Valov et al. / Solid State Ionics 180 (2009) 1463–1470

characteristic changes with the nitrogen concentration. The vacancy concentration increases with increasing nitrogen content, but the total ionic conductivity does not necessarily increase: We found that in general the ionic conductivity decreases with increasing nitrogen content in the lower temperature regime. At temperatures above 1000 °C, the mobility of the nitrogen ions becomes high enough to cause a higher ionic conductivity compared to nitrogen-free YSZ i.e., in fact nitrogen-doped 9.5YSZ becomes a better solid electrolyte — but only at high temperatures. The electronic conductivity as a function of nitrogen content first shows a decrease, which we relate to a decreased mobility of the electronic charge carriers, and then an increase again with increasing nitrogen concentration due to increased electronic state densities near the band edges. At all temperatures and oxygen activities the transference number of electrons was estimated as smaller than te′ < 0.004. The two temperature regions of lower and higher conductivity compared to nitrogen-free YSZ are also distinguished by different activation energies for the ionic conductivity — the latter is high for the low temperature region and low in the high temperature range. The activation energies in the both regions increase with increasing nitrogen concentration. The samples were not chemically stable over the entire temperature range of the measurement cycles (heating and cooling stages), even despite low oxygen activities. A complete re-oxidation and loss of nitrogen were observed at the highest measurement temperatures as confirmed by hot gas extraction analysis. The presence of nitrogen leads to the appearance of a fixed highly negative electrode potential for temperatures at and above 900 °C which is explained by the onset of nitrogen loss coupled to the increasing mobility of nitrogen ions. References [1] J.-S. Lee, M. Lerch, J. Maier, J. Solid State Chem. 179 (1) (2006) 270. [2] A.T. Tham, C. Rödel, M. Lerch, D. Wang, D.S. Su, A. Klein-Hoffmann, R. Schlögl, Cryst. Res. Technol. 40 (3) (2005) 193. [3] J. Wrba, M. Lerch, J. Eur. Ceram. Soc. 18 (1998) 1787. [4] M. Lerch, O. Rahaeuser, J. Mater. Sci. 32 (1997) 1357.

[5] M. Lerch, J. Am. Ceram. Soc. 79 (10) (1996) 2641. [6] M. Lerch, J. Lerch, K. Lerch, J. Mater. Sci. Lett. 15 (24) (1996) 2127. [7] T.-C. Rödel, D. Wang, D.S. Su, M. Lerch, A. Prokofiev, K.D. Luther, W. Assmus, Cryst. Res. Technol. 41 (2006) 950. [8] J.-S. Lee, T.-J. Chung, D.-Y. Kim, Solid State Ionics 136–137 (2000) 39. [9] I. Valov, C. Korte, R.A.D. Souza, M. Martin, J. Janek, Electrochem. Solid-State. Lett. 9 (5) (2006) F23. [10] I. Valov, R.A.D. Souza, C.Z. Wang, A. Börger, C. Korte, M. Martin, K.-D. Becker, J. Janek, J. Mater. Sci. 42 (2007) 1931. [11] M. Lerch, Neue Anionendefizit-Materialien auf der Basis von ZrO2: Synthese, Charakterisierung und Eigenschaften von Ternaeren und Quaternaeren Nitridoxiden des Zirconiums., Habilitationsschrift, 1997. [12] J. Wendel, M. Lerch, W. Laqua, J. Solid State Chem. 142 (1999) 163. [13] T.-J. Chung, J.-S. Lee, D.-Y. Kim, J. Am. Ceram. Soc. 82 (11) (1999) 3193. [14] G. Deghenghi, T.-J. Chung, V. Sergo, J. Am. Ceram. Soc. 86 (1) (2003) 169. [15] M. Kilo, I. Valov, J. Janek, M. Lerch, G. Borchardt, Z. Anorg. Allg. Chem. 630 (2004) 1734. [16] A.T. Tham, C. Rödel, M. Lerch, D. Wang, D.S. Su, A. Klein-Hoffmann, R. Schlögl, Cryst. Res. Technol. 39 (5) (2004) 421. [17] O. Diwald, T.L. Thompson, E.G. Goralski, S.D. Walck, J. John, T. Yates, J. Phys. Chem. B 108 (2004) 52. [18] M. Kilo, M.A. Taylor, C. Argirusis, G. Borchardt, M. Lerch, O. Kaitasov, B. Lesage, Phys. Chem. Chem. Phys. 6 (2004) 3645. [19] M. Taylor, M. Kilo, C. Argirusis, G. Borchardt, I. Valov, C. Korte, J. Janek, T. Roedel, M. Lerch, Defect Diffus. Forum 237–240 (2005) 479. [20] I. Valov, J. Janek, Solid State Ionics 177 (2006) 1619. [21] T. Bredow, Phys. Rev. B 75 (2007) 144102. [22] W. Weppner, J. Electroanal. Chem. 84 (1977) 339. [23] J.-H. Park, R.N. Blumenthal, J. Electrochem. Soc. 136 (10) (1989) 2867. [24] W. Assmus, N. Whippey, Chem. Ing. Technol. 55 (1983) 716. [25] H. Römer, K.-D. Luther, W. Assmus, Cryst. Res. Technol. 29 (6) (1994) 787. [26] R.D. Holmes, A.B. Kersting, R.J. Arculus, J. Chem. Thermodyn. 21 (1989) 351. [27] U. Tinter, H.-D. Wiemhöfer, Solid State Ionics 9–10 (1983) 1213. [28] H.-D. Wiemhofer, Ber. Bunsenges. Phys. Chem. 97 (1993) 461. [29] H.W.S. Lübke, Solid State Ionics 117 (1999) 229. [30] J. Janek, C. Korte, Solid State Ionics 116 (1999) 181. [31] R.D. Shannon, C. Prewitt, Acta Crystallogr. B25 (1969) 925. [32] R. Shannon, Acta Crystallogr. A32 (1976) 751. [33] M. Hendriks, J.T. Elshof, H. Bouwmeester, H. Verweij, Solid State Ionics 146 (2002) 211. [34] J. Nowotny, M. Rekas, T. Bak, Key Eng. Mater. 153–154 (1998) 211. [35] P. Fabry, M. Kleitz, J. Electroanal. Chem. 57 (1974) 165. [36] K. Schindler, D. Schmeisser, U. Vohrer, H.-D. Wiemhöfer, W. Göpel, Sens. Actuators 17 (1989) 555. [37] K. Sasaki, J. Claus, J. Maier, Solid State Ionics 121 (1999) 51. [38] K. Sasaki, J. Maier, Solid State Ionics 134 (2000) 303. [39] H.-D. Wiemhöfer, Ber. Bunsenges. Phys. Chem. 96 (1992) 1646.